child care for families raising children with disabilities ... · in the cost of college tuition...

Child Care for Families Raising Children with Disabilities Page

1

Child Care for Families Raising Children with

Disabilities: The Role of Federal

Policy in Equitable Access

The research reported herein was performed pursuant to a grant from the U.S. Social Security Administration (SSA) funded as part of the Retirement and Disability Consortium. The opinions and conclusions expressed are solely those of the authors and do not represent the opinions or policy of SSA or any agency of the Federal Government. Neither the United States Government nor any agency thereof, nor any of their employees, makes any warranty, express or implied, or assumes any legal liability or responsibility for the accuracy, completeness, or usefulness of the contents of this report. Reference herein to any specific commercial product, process, or service by trade name, trademark, manufacturer, or otherwise does not necessarily constitute or imply endorsement, recommendation, or favoring by the United States Government or any agency thereof.

Molly Costanzo University of Wisconsin-Madison

Katherine Magnuson University of Wisconsin-Madison


2

Abstract Access to affordable and appropriate child care is a crucial employment support for families and

may be particularly salient for families with young children with disabilities. Without access to

quality child care, parents may not be able to find and maintain employment, increasing the

likelihood of economic precarity for families that are already likely to experience economic

hardship. In this study, we use data from the ECLS-B to examine whether child care

arrangements differ by disability status. We find that having an identified disability in childhood,

and particularly IEP receipt, is associated with increased likelihood of attending center-based and

part-time care as well as lower child care costs. We employ a difference-in-difference framework

to examine changes in maternal employment rates at kindergarten enrollment by disability status;

the results of the difference-in-difference analysis are generally not significant, but they do show

a smaller rate of change in maternal employment at kindergarten enrollment for parents of

children with disabilities compared to parents of typically developing peers. These results are

consistent with the conclusion that current federal policies support access to child care for

children with disabilities.

Keywords: child care, disability, policy

JEL Codes: I14, I18, I30, I38,


3

1. Introduction The number of families raising young children with identified disabilities has grown in recent

decades. Using the example of a single diagnostic category, Boyle and colleagues (2011) find a

seventeen percentage-point increase in the proportion of children with developmental disabilities

(defined broadly) born in the United States between 1997 and 2008. This growth means that a

significant number of families are now raising children with disabilities; Halfon and colleagues

(2012) estimate there are currently 1.2 million children under the age of six in the United States

who have a chronic condition that may reflect an underlying disability, representing a prevalence

rate of around 5%.

This increase in the number of children with disabilities has important implications for social

safety net programs; families caring for children with disabilities are at increased risk of

economic insecurity due to increased costs and caregiving needs. Given this increase in

recognized disabilities among children, it is not surprising that the childhood Supplemental

Security Income (SSI) program has been affected, prompting concerns about growth in the

program (Aizer, Gordon, and Kearney 2013; Office of the Assistant Secretary of Planning and

Evaluation 2015). The childhood SSI program is the largest federal program intended

specifically to provide financial support for families raising children with disabilities. Childhood

SSI is a means-tested program that provides modest monthly benefits—$771/month in federal

benefits in 20191—to children who qualify based on strict medical criteria, in addition to income

and asset eligibility requirements2 (SSA 2019).

That support is especially needed because families raising children with disabilities face greater

constraints in balancing employment and caregiving responsibilities, resulting in decreased labor

market participation for these parents (e.g., Parish and Cloud 2006; Stabile and Allin 2012).

Easier access to quality care could help increase labor market participation. It may also have

1 Some states supplement the federal benefit with additional payments. 2 For children who live at home with a parent who is not also an SSI recipient, a portion of the household’s earned, and in some cases unearned, income is considered available to the child through a process called deeming. For more information, see https://www.ssa.gov/ssi/spotlights/spot-deeming.htm.


4

benefits for children; participation in high-quality early care and education settings for children

with disabilities can also support healthy development (Odom, Buysse, and Soukakou 2012).

Yet, the policy and research literature has largely overlooked the role of overlapping and

complementary federal early care and education policies in supporting labor market participation

for parents of children with disabilities.

This study seeks to understand the current landscape of child care access and arrangements for

young children with disabilities in supporting parental employment, indirectly shedding light on

the role of the current set of child care policies in supporting access to care. If families’ need for

child care is met by existing policies and programs and families are able to maintain consistent

employment, then families’ use of public benefits, including childhood SSI benefits, may be

reduced as their household income increases,. Describing the current child care use and

employment patterns of parents who have young children with disabilities can inform

policymakers and researchers about the extent to which the current menu of federal policies is

supporting labor market attachment and thus the potential for economic self-sufficiency for

families raising children with disabilities.

2. Background

Access to affordable and appropriate child care is a crucial employment support for all families,

one that may be particularly salient for families with young children with disabilities, such as

childhood SSI recipients. Indeed, parents raising children with disabilities may face greater

challenges to labor market participation than the general population. Parents of children with

disabilities have a difficult time locating appropriate child care for a host of reasons, including

the limited supply of trained caregivers, perceived discrimination, additional costs, and difficulty

integrating needed services and care (Ceglowski et al. 2009; DeVore and Bowers 2006; Knoche

et al. 2006; Weglarz-Ward and Santos 2018). Thus, access to child care remains difficult, despite

federal policies intended to support access and affordability. Raising a child with a disability is

already associated with economic disadvantage; difficulty accessing early care may exacerbate

economic vulnerability by limiting parental employment (Parish and Cloud 2006).


5

Access to child care may be an overlooked yet key support that could help support families’

financial stability and reduce their use of SSI benefits. In 2017, 1.2 million children under the

age of 18 received childhood SSI benefits (SSA 2018). Children served by the SSI program are,

by definition, more economically disadvantaged than the general population of children and the

overall population of children with disabilities. The number of children served by the program

has increased in recent decades, although it has begun leveling off (Romig 2017). Still, multiple

stakeholders have expressed concern about growth in the program (e.g., American Public Media

2013; Office of the Assistant Secretary of Planning and Evaluation 2015) as well as about how

SSI receipt affects the employment trajectories of recipients’ parents (Deshpande 2016). This

concern is particularly acute for families with very young children; approximately 18% of

childhood SSI recipients in 2018 were under the age of six (SSA 2018). Employment for parents

of these children is likely at least partially dependent on their parents’ ability to find appropriate

child care.

2.1. Review of the Literature

Families with children with disabilities are more likely to live in poverty (Parish and Cloud

2006) and more likely to experience an episode of material hardship than families of typically

developing children (Parish et al. 2008). In part, this finding may be a result of the increased

costs associated with raising a child with a disability. The cost of high-quality child care has

increased for all families over the past decades; in some cases, the increase has outpaced the rise

in the cost of college tuition (Child Care Aware 2017; Laughlin 2013). The economic burden of

care may be particularly difficult for families raising children with disabilities. In qualitative

studies, families raising children with disabilities report high cost of care as a barrier to locating

appropriate, quality care for their children (Glenn-Applegate, Pentimonti, and Justice 2010;

Weglarz-Ward and Santos 2018); some families report that providers charge increased fees to

care for children with disabilities (Ceglowski et al. 2009). These increased fees come on top of

other additional costs, such as higher out-of-pocket healthcare costs, related to raising a child

with a disability (Lukemeyer, Meyers, and Smeeding 2000; Mitra et al. 2017; Shattuck and

Parish 2008). Using the National Survey for Children with Special Health Care Needs, Shattuck

and Parish (2008) estimate that, on average families raising children with disabilities incur an


6

additional $1,000 in child-related costs annually (2018 dollars). This average, however, disguises

the high costs faced by families raising children with severe disabilities; using a more recent

version of the same survey, Lindley and Mark (2010) find that over 20% of families with

children with disabilities had additional child-related costs of more than $1,285 annually (2018

dollars).

In addition to the economic challenges, families with young children with disabilities report

struggling to find care that can accommodate their children’s needs; these challenges include

difficulty coordinating care with other needed services, perceived discrimination and discomfort

on the part of caregivers, and lack of appropriate therapeutic or other support services (Booth-

Laforce and Kelly 2004; Ceglowski et al. 2009; Knoche et al. 2006; Weglarz-Ward and Santos

2018). In interviews, parents report that they perceive their child care options as limited due to

the difficulty of integrating care with outside services (Booth-Laforce and Kelly 2004) or finding

providers who provide supports, such as physical therapy or speech therapy, to meet their

children’s specific needs (Glenn-Applegate, Pentimonti, and Justice 2010).

One major concern that could be addressed by policy is the lack of caregivers trained in early

childhood inclusion. A joint policy statement from the Departments of Health and Human

Services and Education explicitly notes lack of training as a major barrier to inclusion (US

Departments of Health and Human Services and Education 2015). This concern is also borne out

by several studies, from both the parent and provider perspectives (Ceglowski et al. 2009;

Grisham-Brown et al. 2010; Weglarz-Ward, Santos, and Timmer 2019.

Despite these challenges, children with disabilities are enrolled in early care and education

settings. They often start care later in life and may experience a greater number of care

transitions or a greater number of arrangements overall (Booth and Kelly 1999; Ceglowski et al.

2009; DeVore and Bowers 2006; Knoche et al. 2006). Several studies of low-income populations

find no difference by disability status in child care setting, specifically between center-based care

and home-based care (Parish et al. 2005; Wall et al. 2006). Qualitative studies suggest parents

raising young children with disabilities prefer informal care settings (Booth-Laforce and Kelly,


7

2004; Ceglowski et al. 2009). However, more recent analyses of nationally representative data

suggest children with disabilities may use center-based care at higher rates than other children,

and that differences in settings may vary by the child’s age (Costanzo and Magnuson 2019;

Sullivan, Farnsworth, and Sussman-Stillman 2018).

The difficulty of finding appropriate and affordable child care may compromise parents’ ability

to find and keep employment. Labor market attachment for mothers of children with disabilities

is consistently lower than that of mothers of typically developing children, with mothers of

children with disabilities having a decreased likelihood of overall employment and increased

levels of part-time work (e.g., DeRigne and Porterfield 2010; Loprest and Davidoff 2004;

Powers 2001). Evidence suggests that many factors influence the relationship between children’s

health and maternal employment. Several studies find that the definition and severity of

disability impact the magnitude and significance of the estimated effects (e.g., Brandon 2000;

DeRigne, and Porterfield 2010; Lemmon 2015; Loprest and Davidoff 2004; Wasi, van den Berg,

and Buchmueller 2012). Likely as a result of inconsistent diagnosis criteria and groupings, the

evidence is mixed on the type of disability that has the greatest effect on employment, with

various authors pointing to physical disabilities (Lemmon 2015; Wasi, van den Berg, and

Buchmueller 2012) and autism (DeRinge and Porterfield 2010), among others. Severity, however

it may be defined, is also implicated in the magnitude of the effect on maternal employment

(e.g., Brandon 2000; DeRigne and Porterfield 2010). Brennan and Brannon (2005) demonstrated

that the greater the symptomology of the illness, the larger the impact on parental employment

due to less frequent school attendance and less availability of adequate care. Somewhat contrary

to the general findings about severity, Powers (2003), using the 1985–1993 SIPP panels with

three different definitions of disability, found the most expansive (i.e., less severe) definition

rendered the greatest effect on maternal employment, lowering the probability of beginning

employment by 23.8% over a two-year period.

The relationship between childhood disability and parental employment is slightly more nuanced

for SSI beneficiaries. Specifically, SSI payments may allow parents to forego labor market

participation and provide specialized care for their children or SSI receipt may enable parents to


8

pay for specialized care or have greater flexibility in the decision to work or not. The findings in

the literature are decidedly mixed, though most tend to show a negative relationship between SSI

receipt and parental employment (DeRigne and Porterfield 2010). Guldi and colleagues (2018)

find that SSI eligibility—though not necessarily SSI receipt—reduces maternal labor supply

from full to part time. A recent study using SSA administrative data found a similar relationship

with parental earnings; specifically, when a child was removed from SSI, parents responded by

increasing earnings (Deshpande 2016). However, some evidence has found little or no effect on

parental earnings (Duggan and Kearney 2007).

Of central importance to this study is how parental labor market participation is affected in the

early childhood years, given the need for early care and education prior to enrollment in

elementary school. Porterfield (2002) finds a stronger effect of a child’s disability on

employment or hours worked for parents of young children. However, other studies find that the

child’s age does not affect the relationship (Wasi, van den Berg, and Buchmueller 2012).

Notably, studies that consider child’s age often consider it merely as a confounder or conduct

analyses by age group and do not allow the relationship to vary by age; thus, there is no current

evidence as to whether and how the relationship between parental labor supply and childhood

disability varies by child’s age, and, particularly, how it varies from early childhood to

enrollment in kindergarten. This study seeks to fill that gap.

2.2 Policy Context

Like many areas of federal policy in the United States, child care supports are comprised of a

variety of funding streams overseen by a mix of authorities. For this study, we focus on the

constellation of policies intended to influence access to child care for families of children with

disabilities: (1) subsidies available through the Child Care and Development Fund (CCDF), (2)

Head Start and Early Head Start, and (3) the Individuals with Disabilities in Education Act

(IDEA). The policy mechanisms differ; some are intended to lower economic costs, some to

guarantee access, and some to do both. In this study, we are interested broadly in whether this

package of policies creates access to care and do not focus specifically on the role of any one


9

policy or mechanism. Thus, we briefly describe the policies to further motivate the analysis and

provide context for study design, findings, and discussion.

The Child Care and Development Fund (CCDF) is a block grant program that provides child care

subsidies to low-income parents who are employed. Recent reauthorizations require states to

give priority for funding to children with disabilities (US Department of Health and Human

Services 2016) and develop strategies to increase the availability of high-quality care for children

with disabilities (US Department of Health and Human Services 2015), though some states were

already prioritizing children with disabilities prior to these requirements (US Department of

Health and Human Services 2003). One recent study using the Early Childhood Longitudinal

Study, Birth Cohort (ECLS-B) found that subsidies increased use of nonparental care,

particularly center-based care, for children with special healthcare needs (Sullivan, Farnsworth,

and Sussman-Stillman 2018).

Like CCDF, which may offset costs for families, Head Start and Early Head Start are required by

statute to set aside 10% of their enrollment slots for young children with disabilities (45 CFR

§1308). IDEA may allow access to care for children who qualify both by ensuring receipt of

required services and by providing necessary care at no cost to the family. IDEA mandates

inclusion for children with disabilities in all education settings, including early childhood and

child care (US Department of Education 2016). IDEA also provides funding for nonparental care

for children with disabilities for the portion of the day when the child receives services,

defraying some of the cost of care (U.S. Library of Congress 2016). Unlike subsidies and Head

Start and Early Head Start, which are income limited, children from households across the

income distribution are entitled to services via IDEA.3

Though we focus on access to care for children with disabilities, the general child care

environment for all families provides important context for the study. In particular, the costs of

3 There is essentially no empirical work examining interactions between child care programs and childhood SSI, though the program policies are explicit about some of the eligibility relationships. For example, receipt of SSI qualifies a child for enrollment in Head Start or Early Head Start (Office of Head Start 2019). SSA policy explicitly states that subsidies for child care received through CCDF should not be included as countable income in determining SSI eligibility (20 CFR 416.1102, 416.1103, and 416.1124(b)).


10

child care are high for all families, and many parents of typically developing children also

struggle to locate appropriate, affordable care (Laughlin 2013). Some of the policies intended to

support children with disabilities also apply to children in the general population. Specifically,

both Head Start and child care subsidies are available to typically developing children from low-

income families. However, use of subsidies is limited; estimates of utilization range from 7 to

34% of income-eligible families (Forry, Daneri, and Howarth 2013). In recent years, interest in

implementing public preschool programs has grown. In 2018, 44 states provided publicly funded

preschool for some children for some portion of the day; still, a limited number of children enroll

in public preschool—approximately one-third of all four-year-olds and just under 6% of all

three-year olds (National Institute for Early Education Research 2019). Availability of high-

quality care for very young children—infants and toddlers—has lagged that of preschoolers

(Lally et al. 2003).

This study examines the extent to which parents of children who have disabilities have differing

child care arrangements and parental employment patterns compared with parents of typically-

developing children. Though the research literature has focused on parental employment overall

and has provided some indication of the difficulty families have in accessing early care and

education, evidence about access to early care and parental employment in the early years is

currently lacking. In particular, little is known about the efficacy of the package of policies

including CCDF, Head Start and Early Head Start, and IDEA in fulfilling its goal of ensuring

free and appropriate access to early education and care for young children with disabilities. Nor

is it known whether this access is, in turn, supporting parental employment. One recent study,

using the ECLS-B, finds a relationship between subsidy eligibility and child care use (Sullivan,

Farnsworth, and Sussman-Stillman 2018). However, no current national study attempts to

understand the effect of the current constellation of policies related to early care for children with

disabilities on parental employment and economic stability. Understanding whether the current

landscape of complementary policies and programs that, like the childhood SSI program, are

intended to support families of children with disabilities is salient for policy makers.

3. Data and Methods


11

We use data from the Department of Education’s ECLS-B. The ECLS-B is representative of

children born in the United States in 2001 and designed to provide information about children’s

health and development from birth through enrollment in kindergarten. Data are collected from

parent interviews, which were conducted at four or five time points, depending on the child’s

year of kindergarten enrollment. Parents of 10,700 children4 born in 2001 were interviewed in

the first interview (wave 1), around the child’s nine-month birthday. The second interview (wave

2) was intended to coincide with the child’s second birthday; data were collected from parents of

9,850 children in this wave, as well as in the third interview (wave 3, or the preschool wave),

which occurred during academic year 2005–2006. Parents were also interviewed during the

2006–2007 academic year (wave 4, or the kindergarten 2006 wave), which is the year 75% of the

sample began kindergarten. Because states have different age cutoffs for kindergarten enrollment

and because age eligibility may not follow the calendar year, some children did not begin

kindergarten until the following school year. Parents of these children (approximately 1,550)

were interviewed both during the first kindergarten wave (2006–2007) and again the following

academic year (wave 5), when their children enrolled in kindergarten for the first time (Snow et

al. 2009). We include children who entered kindergarten in both waves and control for year of

kindergarten entrance in analyses. Interviews for the last three waves of data collection could

have occurred at any time between September and March of the academic year of collection; we

also account for this timing in our analysis.

The ECLS-B data are well suited for the current study because they are nationally representative

as well as longitudinal, alleviating endogeneity concerns that may arise with cross-sectional data.

The survey collected detailed information about children’s health and development and about

parental employment. In addition, the ECLS-B includes a relatively large sample of children

with disabilities, thereby relieving some concerns about statistical power or sample size in other

datasets. In this study, we use data from the parent interviews, which contain detailed

information about (1) diagnosed health conditions, (2) child care arrangements and use, (3)

receipt of Individualized Education Programs (IEPs),5 and (4) parental employment.

4 As required by the Department of Education, we present only weighted proportions and round all sample sizes to the nearest 50. 5 An IEP is the document that specifies the specialized services a child is eligible for under IDEA.


12

The child’s mother was the respondent in a majority of the cases across waves. We limit our

analytic sample to children whose mother was present in the household and completed the survey

in the focal child’s kindergarten year, which could have been either 2006–2007 or 2007–2008

(N=6,900).6 To reduce bias, we also exclude children who were diagnosed with a disability

during their year of kindergarten enrollment from our difference-in-difference analysis, leaving

an analytic sample of 6,700 children. When presenting descriptive statistics, we use weights

suggested by the Department of Education to adjust for the complex sampling design and

nonresponse. Because our models account for endogenous sampling concerns, we present

multivariate results using unweighted models with robust standard errors; however, results are

robust to the inclusion of weights (Solon, Haider, and Wooldridge 2015.

In addition to the main analysis, we also used multiple imputation with chained equations

(MICE) to account for missing data and disproportionate nonresponse by disability status. As a

robustness check, we conduct all analyses on the analytic sample with imputed data for missing

responses. We also run models on the full ECLS-B sample (N=10,700) with imputed data for all

missing data, including missing waves.7

3.1. Measures

Disability. The key independent variable is the focal child’s disability status. Parents were asked

in each wave of interviews whether the focal child had been diagnosed with any of a variety of

health conditions. The conditions varied by interview wave8 but generally included measures of

intellectual disability, developmental delays, autism, issues with mobility, mental health

diagnoses, speech and language conditions, blindness, difficulty hearing, and other conditions.

6 Approximately 100 cases do not have a household mother when the child is in kindergarten; these are excluded from the sample. A very small number of cases did not enroll in kindergarten by the 2007 wave (Snow et al. 2009); these cases are also excluded from our sample. 7 Additional information on our MICE models, including analytic results, is available by request. 8 Appendix A contains a detailed list of the conditions included in each wave of interviews.


13

Table 1: Proportion of ECLS-B Sample That Meets Various Disability Definitions

Disability Measure Operationalization Wave Weighted

Proportion Population Size Unweighted N

General Disability Binary indicator for

parents who responded affirmatively to the question, "Has your child been diagnosed with any of the following conditions . . .?"

Wave 1 6.3% 247,856 600

Wave 2 9.7% 379,620 900

Wave 3 13.9% 541,618 1,100

Wave 4 15.5% 604,177 1,200

Wave 5 21.7% 259,489 500

Multiple Diagnoses Indicator for parents who responded that the child had more than one diagnosis in the conditions listed

Wave 1 1.2% 45,398 150

Wave 2 2.9% 113,457 350

Wave 3 5.0% 196,227 450

Wave 4 5.2% 202,832 450

Wave 5 7.6% 90,877 200

Continuing Disability Measure, Prior to Kindergarten

Once a parent answers yes to the disability question, the parent continues to report the child's disability through all remaining waves

Waves 1–3/4

8.4% 328,742 700

& All Waves All 14.4% 562,384 1100

Disability in any Wave, Prior to Kindergarten &

A parent responded that the child had a disability in any interview wave

Waves 1–3/4

23.2% 902,989 1,850

All Waves All 29.2% 1,141,895 2,250

Disability in any Wave, Including Asthma

A parent responded that the child had a disability, including asthma, in any interview wave

All 43.8% 1,712,151 3,300

Child's Health is Fair or Poor

A parent rated the child's health fair or poor on the 5-category self-rated health scale

Wave 1 2.3% 89,987 200

Wave 2 2.2% 86,099 200

Wave 3 2.6% 100,046 200

Wave 4 2.0% 78,944 200

Wave 5 2.1% 25,111 50

IEP, IEP before Kindergarten

The child has an Individualized Education Program in place

Wave 3 5.1% 197,395 450

Wave4 5.4% 212,248 500

Wave 3/4 5.6% 218,894 500

Received Special Services

The parent reported that the child received intervention services in the previous month, including physical therapy, occupational therapy, speech and language therapy, or other support services

Wave 1 2.5% 97,776 350

Wave 2 2.5% 99,377 350

Wave 3 3.2% 125,199 250

Wave 4 5.1% 198,363 400

Wave 5 8.3% 99,080 200


14




Received Early Intervention Services Prior to Kindergarten

The parent reported that the child received Early Intervention Services, in any location

Waves 1–3/4

6.1% 239,058 600

Specific Diagnoses

Communication Related

Specific measure of child having difficulty hearing, talking, or other communication-related condition (including speech and language delays)

Wave 1 0.7% 28,952 50

Wave 2 4.5% 175,950 750

Wave 3 10.4% 405,467 1100

Wave4 8.7% 341,343 700

Wave 5 10.2% 19,482 250

Waves 1–3/4

11.9% 464,414 1050

Intellectual Disability Specific measure of intellectual disability (e.g., Down Syndrome, Turner Syndrome, intellectual disability)

Wave 1

—

Wave 2 0.4% 13,697 —

Wave 3 0.2% 7,815 50

Wave4 0.4% 14,459 50

Wave 5 0.4% 5,142 —

Waves 1–3/4

0.6% 25,040 50

Emotional or Mental Health

Specific measure of emotional or mental health conditions (e.g., oppositional defiant disorder, ADD, ADHD)

Wave 3

4.3% 168,184 400

Wave 4 5.1% 199,316 400

Wave5 7.8% 93,221 200

Waves 3/4 6.7% 262,055 550

Autism Specific measure of autism Wave 3 0.9% 34,370 100

Wave 4 1.1% 44,153 100

Wave5 1.0% 11,944 50

Waves 3/4 1.2% 46,498 100

Physical/Orthopedic Specific measure of

challenges with mobility, limbs, or a diagnosis of spina bifida

Wave 1 1.3% 496,885 150

Wave 2 0.6% 22,698 100

Wave 3 2.6% 101,618 300

Wave4 2.1% 82,051 200

Wave 5 2.5% 29,894 100

Waves 1–3/4

4.1% 160,803 400

Chronic Condition Specific measure of chronic conditions, including heart

Wave 1 2.7% 104,463 200

Wave 2 2.3% 91,187 200


15




conditions, epilepsy, diabetes, or anemia

Wave 3 2.6% 102,799 200

Wave4 3.2% 124,279 250

Wave 5 3.6% 43,047 100

Waves 1–3/4

6.0% 234,669 500

Congenital Syndromes, Identifiable at Birth

Specific measure of Down Syndrome, Turner Syndrome, or Spina Bifida

All Waves 0.2% 7824.896 —

Source: ECLS-B

One of the major methodological considerations with this study, as with all studies concerned

with disability, is the imprecise and heterogenous nature of the disability construct, as outlined in

Table 1. The table also showcases the consequential—and challenging—nature of measuring

disability in this and other common data sets. At the high end of prevalence estimates, using the

broadest definition, which includes measures for children diagnosed with asthma and children

who require vision correction with glasses, 44% of the sample has a disability diagnosis. The

proportion of the sample in each wave with a disability begins at about 6% at the nine-month

wave and increases to 16 to 22% in the kindergarten waves; these are children whose parents

reported that they had been diagnosed with a specific health condition in the previous year.

About 23% of the sample has a disability in at least one wave prior to kindergarten, even if they

do not report having a disability in later waves; 8% have a disability continuously once they

report having a disability (and have a diagnosis prior to kindergarten enrollment). The proportion

of children who receive intervention services or who have IEPs is substantially smaller than the

proportion who report having a disability. Approximately 6% of the sample had an IEP in place

prior to kindergarten enrollment and also reported receiving early intervention services. We also

examined the distribution of some of the disability categories of the sample; communication-

related diagnoses are the largest group represented (12% with a diagnosis before kindergarten)

and intellectual disability is the least represented (less than 1%). Note that the list is not

exhaustive and thus the estimates do not total the overall measures.


16

We present results using three binary measures of disability, as measured before the child’s

enrollment in kindergarten: (1) a continuing measure of disability, (2) a measure of disability in

any wave, and, for some models, (3) an indicator for having an IEP. We exclude diagnoses of

asthma and difficulty seeing in the first two measures for a few reasons. First, these conditions

are often relatively manageable and may not require time-intensive care from a parent or child

care provider. Second, neither is explicitly covered under applicable federal policies. Finally, less

than 1% of children with asthma or difficulty with vision report receiving regular special

services. The first measure—continuing disability—indicates that once the parent reports that the

child has a disability, the disability continues to be reported in each subsequent interview. We

restrict this measure to children whose disabilities were identified prior to kindergarten

enrollment in an attempt to disentangle diagnostic and parental employment timelines. The

second measure—disability in any wave prior to kindergarten enrollment—is intended to be a

broad measure of whether a parent ever reports the child has a disability in any wave of data

collection. Though it is likely this measures biases the estimate of the number of children with

disabilities upwards, the parent survey asks whether a child has been diagnosed since the

previous survey, so it is also possible that the previous measure of disability does not capture

parents who no longer report the condition in subsequent interviews based on the wording of the

survey.

We select these measures for a few reasons. First, given the difficulty in measuring disability,

taken together, the measures may help bound the estimated impact of having a child with a

disability on child care access and parental employment. If the definitions exist on a spectrum of

severity, children who are identified as having a continuing disability may represent an upper

bound of the effect of having a child with a disability while the other measure—disability in any

wave—may offer a lower bound; the second measure may capture children with less severe

conditions or temporary conditions that may be less likely to influence child care access.

Additionally, these measures may estimate the effect of two different aspects of disability. While

the first measure—continuing disability—captures children with an ongoing health condition, the

second includes children who may have experienced a health shock or whose health condition

may have improved. We use IEP receipt in our child care analyses because of its policy


17

relevance; children with IEPs are receiving services under IDEA. Thus, estimates for this

measure help us understand the role of IDEA in child care access and parental employment.

Child Care Arrangements. One of our primary outcomes of interest is child care arrangements.

We employ multiple of measures of child care use. First, to determine whether a child is in

regular nonparental child care, we use a binary indicator for parental report of regular

nonparental care at least ten hours each week. We are also interested in understanding patterns in

the length of time children are in nonparental care each week. We use a continuous measure of

the number of hours each week a child is typically in nonparental care, across all arrangements.

We also use binary measures for full-time care, which includes children in nonparental care for

more than 30 hours each week, and part-time care, which includes children in care at least ten

hours but less than or equal to 30 hours each week. Given public funding for and emphasis on

access to center-based care settings, investigating patterns of arrangements is also important to

understand how policy may be creating access. Thus, we include measures of the child’s primary

care arrangement, defined as the arrangement in which the child spends the most time each

week. We define the arrangements as informal or relative care, if the child is in care by a

relative, regardless of the location of the care, or a nonrelative in the child’s home; home-based

care if the child is in care with a nonrelative at a location other than the child’s own home (e.g.,

in-home providers); and center-based if a child is in care at a child care center or preschool. We

also include a measure for whether the child is ever enrolled in Head Start. Children in different

arrangements for equal time default to center-based care first (due to its policy relevance), then

home-based care. Finally, we look at the cost of care for families, both annually and hourly. In

some models, we also use a binary indicator for whether a child is receiving a child care subsidy;

the indicator is based on a parent’s report that an institution of some kind is helping with child

care payments, or a parent’s report of paying no fee for home-based or center-based care (with

the exception of Head Start). Thus, this indicator is not a precise measure of subsidies received

under CCDF.

Employment. Another key dependent variable is parental employment, with a focus on maternal

employment specifically. We use several measures of employment in this analysis. The first is a


18

binary measure of whether a child’s mother is employed at the time of the survey, regardless of

the number of hours worked. We also include binary indicators for full-time employment (at

least 35 hours per week) and part-time employment. In addition, we include measures of the

continuous hours worked by the household mother respondent in the last week and total number

of hours worked by all parents in the household.9 In some models, we also include a binary

indicator for household father employment. We focus on employment outcomes at two time

points: employment during the preschool wave (wave 3) and employment during the child’s first

year of kindergarten enrollment. Since the year of kindergarten enrollment varies across the

sample, this measure could be employment at wave 4 for children who first enrolled in

kindergarten in wave 4 or employment at wave 5 for children who first enrolled in wave 5. We

account for these differences with a binary indicator for year of entrance.

Covariates. We include covariates that are likely related to either child care and disability or

parental labor supply and disability. For the child care models, these covariates include

indicators of the child’s race and ethnicity (white, black, Latinx, or other race) and sex and a

four-category indicator for household income-to-poverty ratio, adjusted for household size

(below 100% of the 2005 federal poverty threshold, between 100% and 200% of the threshold,

between 200% and 400% of the threshold, and above 400%). Models predicting parental

employment use the mother’s race instead of the child’s, a control for whether the mother was

employed in the 12 months prior to the child’s birth, a measure for whether the child is

homeschooled or attending half-day kindergarten, the year the child entered kindergarten, and

the quarter in which the interview occurred. All models include a categorical indicator of highest

parental education level, an indicator for single-mother households, the number of children under

six in the household, the number of other children in the household, the number of nonparental

adults in the household, an indicator for poor or fair maternal health status, and an indicator for

urbanicity. Each model also includes state fixed effects.

Ideally, we would also account for employment in our child care analyses and household

economic status in our employment analyses. However, there are clear endogeneity concerns.

9 These analyses are restricted to households in which the household mother was the survey respondent.


19

Households are unlikely to make child care and parental employment decisions independently,

and the causal direction of the relationship is not clear. Similarly, though household economic

status likely drives parental employment behavior, it also directly influences household income.

We use income-to-poverty ratio in wave 3 when considering child care arrangements at the same

time point. For employment outcomes, we use income-to-poverty ratio at wave 1, which is

conceptually less problematic than income from the same time point. We also include a measure

of parental education level as a proxy for socioeconomic status. Additionally, we conduct

robustness checks including relevant variables.

3.2 Statistical or Analytic Methods

To describe child care use for children with disabilities, we first use multivariate regression to

predict child care use by disability status, holding constant other relevant demographic and

household characteristics. We focus our main analysis on the preschool wave (wave 3, which

occurred the year the child turned four), because this is the age at which the majority of children

are enrolled in early care or education.10 We use OLS models to test the association between

disability status and hours in care and cost of care. We use logistic regression to predict regular

nonparental child care (for at least ten hours each week), full-time child care (defined as greater

than or equal to 30 hours/week), and part-time child care (more than ten but fewer than 30 hours/

week). We also examine the type of child care use, with logistic regressions predicting center-

based care, home-based care, and Head Start, as well as a multinomial logistic regression

comparing the outcomes of no regular nonparental care, informal care (combining relative care

and home-based care), or center-based care.

Our analytic models are below, where ℓ 𝑐𝑐ℎ𝑖𝑖𝑖𝑖𝑖𝑖 𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐 𝑖𝑖𝑑𝑑 is the log odds of the binary child care

variables, β1Disability is the indicator for childhood disability, and β2X is a vector of covariates

described above. The multinomial model is represented by Equations 1 and 2 below, where 𝑙𝑙𝑙𝑙𝑙𝑙

(𝜋𝜋𝑖𝑖

(𝐶𝐶𝐶𝐶/𝐼𝐼𝐶𝐶)

𝜋𝜋𝑖𝑖𝑁𝑁𝐶𝐶/𝐶𝐶𝐶𝐶 ) represents the odds ratio of the outcomes of informal care (IC) and center-based care

(CC) compared to the base outcome of no care (NC) in Equation 1, and the outcomes of informal

10 Results for earlier waves of care are available in Appendix B. More information is available upon request.


20

care compared to the base outcome of center care in Equation 2 for an individual, i; 𝛽𝛽1(𝐼𝐼𝐼𝐼)

disabilityi is the key predictor; and 𝛽𝛽2(𝐼𝐼𝐼𝐼)𝜒𝜒 𝑖𝑖 is the vector of covariates.

Equation 1: ℓ𝑐𝑐ℎ𝑖𝑖𝑖𝑖𝑖𝑖 𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐 𝑖𝑖𝑑𝑑 = β0 + β1Disability + β2X + ε

Equation 2: 𝑌𝑌ℎ𝑜𝑜𝑜𝑜𝑐𝑐𝑜𝑜/𝑐𝑐𝑜𝑜𝑜𝑜𝑐𝑐 = β0 + β1Disability + β2X + ε

Equation 3: 𝑙𝑙𝑙𝑙𝑙𝑙 (𝜋𝜋𝑖𝑖

(𝐻𝐻𝐶𝐶/𝐶𝐶𝐶𝐶)

𝜋𝜋𝑖𝑖𝑁𝑁𝐶𝐶 ) = 𝛼𝛼

(𝐻𝐻𝐶𝐶𝐹𝐹𝐶𝐶 ) + 𝛽𝛽1(𝐻𝐻𝐶𝐶𝐶𝐶𝐶𝐶) disabilityi + 𝛽𝛽2(𝐻𝐻𝐶𝐶𝐹𝐹𝐶𝐶)𝜒𝜒 𝑖𝑖 + 𝜀𝜀𝑖𝑖

Equation 4: 𝑙𝑙𝑙𝑙𝑙𝑙 (𝜋𝜋𝑖𝑖

(𝐻𝐻𝐶𝐶)

𝜋𝜋𝑖𝑖𝐶𝐶𝐶𝐶 ) = 𝛼𝛼(𝐻𝐻𝐼𝐼) + 𝛽𝛽1(𝐻𝐻𝐼𝐼) disabilityi + 𝛽𝛽2(𝐻𝐻𝐼𝐼)𝜒𝜒 𝑖𝑖 + 𝜀𝜀𝑖𝑖

We use a difference-in-difference framework (DD) to examine the difference in the rate of

change of maternal employment at kindergarten enrollment between parents of children with a

disability and parents of typically developing children. The DD framework compares a

“treatment” group to a control group, comparing outcomes before and after a specific

intervention or policy event. If the major assumptions of the method hold, using DD

methodology can provide a stronger foothold in causality than regression alone. In addition, the

intuition behind DD offers a clear interpretation of results. Theoretically, unlike early care and

education, which may have barriers to access for all children and potentially for children with

disabilities in particular, access to kindergarten is free and universal. Given this change at the

time of kindergarten enrollment, parental employment is likely to increase for all children. If

current policies are not supporting access to child care for children with disabilities, the effect

will be larger for parents of children with disabilities; that is, the increase in parental

employment rates will be greater for this group. If, however, current policies are supporting

access to child care for these families, the analysis would show a similar effect on parental

employment at kindergarten enrollment for both groups, or potentially a smaller change for

parents of children with disabilities.

We compare parental employment outcomes for children with disabilities to typically developing

children. In our main DD analyses, we use data from the latter two waves (or three if the child

enrolled in kindergarten in 2007) and compare outcomes from both groups from parent

interviews pre-kindergarten to kindergarten enrollment. Our analytic model is outlined in


21

Equation 5, where Yit is a measure of maternal employment for individual i and time t. The

dummy variable 𝐵𝐵1𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝑙𝑙𝐷𝐷𝐷𝐷𝐷𝐷it is an indicator for a child with a disability, 𝐵𝐵2𝐾𝐾𝐷𝐷𝐾𝐾𝐾𝐾𝐾𝐾𝐾𝐾𝑙𝑙𝐷𝐷𝐾𝐾𝐷𝐷𝐾𝐾𝐾𝐾it is

an indicator for kindergarten enrollment, and 𝛿𝛿(𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝑙𝑙𝐷𝐷𝐷𝐷𝐷𝐷 ∗ 𝐾𝐾𝐷𝐷𝐾𝐾𝐾𝐾𝐾𝐾𝐾𝐾𝑙𝑙𝐷𝐷𝐾𝐾𝐷𝐷𝐾𝐾𝐾𝐾)it is the DD

estimate for the effect of kindergarten enrollment on maternal employment rates for mothers of

children with disabilities. 𝛽𝛽𝛽𝛽it is the vector of covariates including maternal health status,

parental education level, and household and demographic variables. We also include controls for

half-day kindergarten and year of kindergarten enrollment, as well as state fixed effects.

Equation 5: Yit = β0 + β1𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝑙𝑙𝐷𝐷𝐷𝐷𝐷𝐷i + β2𝐾𝐾𝐷𝐷𝐾𝐾𝐾𝐾𝐾𝐾𝐾𝐾𝑙𝑙𝐷𝐷𝐾𝐾𝐷𝐷𝐾𝐾𝐾𝐾t +

𝛿𝛿(𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝑙𝑙𝐷𝐷𝐷𝐷𝐷𝐷 ∗ 𝐾𝐾𝐷𝐷𝐾𝐾𝐾𝐾𝐾𝐾𝐾𝐾𝑙𝑙𝐷𝐷𝐾𝐾𝐷𝐷𝐾𝐾𝐾𝐾)it + 𝛽𝛽𝛽𝛽it + 𝜀𝜀

DD requires additional assumptions beyond the general assumptions for all multivariate

regression analyses to obtain unbiased estimates. In particular, DD relies on the parallel trends

assumption, which, in this case would mean that maternal employment rates for children with

and without disabilities follow the same time trends in absence of treatment. That is, the

difference in the level of maternal employment between the two groups would remain relatively

stable in the absence of kindergarten enrollment. We are able to provide some evidence that this

assumption is met. Additionally, the DD framework assumes that the composition of each group

cannot be altered by the intervention itself. In this case, whether or not a child has an identified

disability cannot be determined by enrollment in kindergarten. In order to assure that this

assumption is not violated, we include only children who have been identified as having a

disability prior to kindergarten enrollment. Results are robust when we also include children who

are identified as having a disability in kindergarten.

4. Results 4.1. Main Results

There are some differences in key characteristics by disability status (Table 2). In particular,

children with disabilities are disproportionately likely to live in households with a single parent

and with a mother reporting fair or poor health. As expected, there are also differences in birth


22

weight status as well as likelihood of the household ever having received SSI or SSDI.11 Across

all measures, boys and children who are white are more likely to be identified as having

disabilities. Differences in kindergarten enrollment are apparent by homeschool status and the

increased likelihood for children with disabilities to begin kindergarten in the second year of

enrollment covered by the survey.

Table 2: Sample Summary Statistics, Selected Characteristics by Disability Type,

All Any Disability,

Pre-K Continuing Disability,

Pre-K Yes No SS Yes No SS N= 6900 5100 4750 700 6250

Male 51.2% 58.4% 49.0% *** 64.7% 50.0% ***

Child race/ethnicity White 53.7% 59.0% 52.2% *** 59.1% 53.2% ** Black 13.9% 13.3% 14.1% 12.6% 14.0%

Latinx 25.1% 21.6% 26.2% *** 21.1% 25.5% * Other race 7.2% 6.1% 7.6% * 7.1% 7.2%

Low birth weight 7.5% 11.5% 6.3% *** 10.7% 7.2% ***

Single parent HH (wave 3) 21.9% 25.1% 21.0% ** 24.9% 21.6% Number of children in HH ≤ 5 (wave 3) 1.6 1.6 1.6 1.7 1.6 Number of siblings, total (wave 3) 1.4 1.4 1.4 1.4 1.4 Number of nonparental adults in HH (wave 3) 0.38 0.41 0.37 0.46 0.37 * HH income, mean (wave 1) $50,720 $49,773 $51,019 $49,829 $50,802 Family income as % FPL (wave 3)

< 100% FPL 23.5% 23.7% 23.5% 24.6% 23.4% 100–199% FPL 28.8% 28.8% 28.8% 27.3% 28.9% 200–399% FPL 24.3% 26.4% 23.7% 27.0% 24.0%

400%+ FPL 23.4% 21.2% 24.1% * 21.2% 23.6% HH income ever below FPL (waves 1–3) 37.0% 36.9% 37.1% 40.0% 36.8% HH income ever below 200% FPL (waves 1–3) 63.8% 64.5% 61.4% 65.8% 61.8% HH received SSI/DI benefits 8.9% 14.7% 7.1% *** 22.0% 7.7%

***

HH ever received SSI/DI (since child's birth) 9.5% 11.9% 8.7% ** 10.4% 9.3% ** Parents' highest level of education (wave 3)

High school or less 33.8% 33.8% 33.8% 32.8% 33.9% Some college 33.3% 35.7% 32.6% * 36.9% 32.9%

11 The survey does not differentiate which member of the household is eligible for SSI/SSDI; thus, it is not clear whether the receipt of benefits is due to the health of the child or that of another household member.


23

Table 2: Sample Summary Statistics, Selected Characteristics by Disability Type,

All Any Disability,

Pre-K Continuing Disability,

Pre-K College or higher 32.9% 30.5% 33.6% 30.0% 33.3%

Urbanicity (wave 3) Urban, Large 71.7% 68.6% 72.6% ** 67.5% 72.1% * Urban, Small 11.7% 12.8% 11.4% 13.3% 11.6%

Rural 16.6% 18.5% 16.0% 19.2% 16.3% Census region (wave 3)

Northeast 16.3% 15.1% 16.7% 15.2% 16.4% Midwest 21.7% 25.5% 20.1% *** 26.9% 21.2% **

South 38.0% 38.5% 37.8% 35.6% 38.2% West 24.0% 21.0% 25.0% ** 22.3% 24.2%

Mother Employed 12 months before birth 71.9% 72.1% 71.8% 72.3% 71.8% First enrolled in K in wave 4 (2006) 73.0% 68.0% 74.4% *** 66.3% 73.6%

***

Homeschooled for kindergarten 1.8% 1.0% 2.0% *** 1.1% 1.8%

IEP 5.1% 21.5% 0.1% *** 41.1% 1.7% ***

*** p<0.01, ** p<0.05, * p<0.1 Source: Authors’ calculations using ECLS-B data

Patterns of child care use suggest an increased use of nonparental care overall as children age

and a significant shift to center-based care from home-based care at the preschool wave (Table

3). This trend is not surprising, given the shift in availability and reduction in costs of center-

based care for preschoolers compared to infants and toddlers. Some differences by disability

status are evident in the table; most of these differences emerge in the preschool wave. Children

with disabilities are significantly more likely to be in any regular nonparental care, more likely to

be in part-time care, and more likely to be in full-time care than typically developing peers.

Children with disabilities are also more likely to be enrolled in center-based care and, by some

measures, less likely to be in relative or informal care. Disability status is also associated with

some sort of subsidy for child care use. Perhaps as a result, parents of children with disabilities

are paying, on average, less for care, both annualized and hourly. Finally, children with

disabilities began nonparental care of all types and center-based care specifically at later ages

compared with other children.

Table 3: Child Care Arrangements by Disability Status (Bivariate)

Wave All Any Disability, Pre-K Continuing Disability, Pre-K Yes No SS Yes No SS

Reg

ula

r Avg Hours in Care/Week

1 15.96 15.99 15.94 14.45 16.10 2 16.00 15.80 16.04 13.85 16.19 **


24

Table 3: Child Care Arrangements by Disability Status (Bivariate)

Wave All Any Disability, Pre-K Continuing Disability, Pre-K 3 22.63 23.58 22.33 24.95 22.41 **

Regular Child Care

1 42.8% 42.5% 42.9% 39.6% 43.1% 2 44.2% 43.4% 44.4% 40.6% 44.5% 3 65.5% 69.7% 64.2% *** 75.6% 64.6%

***

PT

vs

FT

Car

e

PT Care (10–29

hrs/wk)

1 12.5% 12.5% 12.5% 12.5% 12.5% 2 12.5% 12.3% 12.5% 15.9% 12.2% 3 25.2% 27.7% 24.5% 32.2% 24.6% **

FT Care (>/=30 hrs/wk)

1 30.3% 30.0% 30.4% 27.1% 30.6% 2 31.7% 31.1% 31.9% 24.7% 32.4%

***

3 40.3% 42.0% 39.6% 43.5% 40.0% *

Pri

mar

y C

are

Arr

ange

men

t T

ype

Relative Care/Informa

l Care

1 23.8% 23.2% 24.0% 22.6% 23.9% 2 18.9% 17.1% 19.5% 16.3% 19.2% 3 13.4% 10.7% 14.3% *** 12.0% 13.6%

Center-Based Care

1 8.0% 7.8% 8.0% 7.3% 8.0% 2 13.9% 13.5% 14.1% 14.7% 13.9% 3 46.8% 52.5% 45.1% *** 56.9% 45.9%

***

Home-Based

Care

1 11.1% 11.5% 11.1% 10.0% 11.3% 2 11.4% 12.8% 11.0% 9.6% 11.6% 3 5.2% 6.4% 4.8% ** 6.8% 5.1%

Pol

icy-

Rel

evan

t

Head Start 3 3.6% 4.5% 3.3% 4.0% 3.5% IEP 3 5.1% 21.5% 0.1% *** 41.1% 1.7%

***

Child Care Subsidy

1 6.2% 6.0% 6.3% 6.2% 6.3% 2 8.2% 8.8% 8.0% 10.4% 7.9% 3 20.0% 27.3% 17.7% *** 37.6% 18.4%

***

Cos

t of

Car

e

Cost of Care (Annual)

1 $2,695.66 $2,636.66 $2,711.45 $2,519.17 $2,709.89 2 $3,209.63 $3,241.84 $3,200.54 $2,904.37 $3,234.62 3

$2,529.44 $2,216.03 $2,627.95 ** $1,829.34 $2,597.88 ***

Cost of Care (Hourly)

1 $ 1.92 $ 1.62 $ 2.00 *** $ 1.61 $ 1.94 * 2 $ 2.17 $ 2.18 $ 2.17 $ 2.00 $ 2.19 3 $ 2.25 $ 1.88 $ 2.36 *** $1.39 $ 2.33

***

Ear

liest

Age

(M

onth

s)

All Types — 12.17 13.27 11.83 * 14.55 11.94 **

Center-Based Care — 28.25 29.47 27.86 ** 30.50 28.01 **

*** p<0.01, ** p<0.05, * p<0.1 Source: Authors’ calculations using ECLS-B data


25

To further unpack the relationship between disability status and child care use in the preschool

wave, we conducted multivariate regression models predicting child care use and child

care type by disability status while controlling for relevant demographic and household

characteristics (Table 4). Overall, results indicate that both primary measures of disability are

significantly associated with increased likelihoods of regular care use for at least ten hours each

week, part-time care, and center-based care, with larger magnitudes for children with continuing

disabilities than for children with disabilities in any wave. Children with a continuing disability

are 9% more likely to be enrolled in regular care, 8% more likely to be enrolled in part-time care,

and 11% more likely to be enrolled in center-based care than typically developing peers.

Children with a disability in any wave before kindergarten are 4% more likely to be in any

regular care, 4% more likely to be in part-time care, and 6% more likely to be in center-based

care than typically developing peers.

The policy-relevant measure of disability—whether the child has an IEP—is strongly associated

with increased use of care overall (children with an IEP are 18% more likely to be in care than

children without an IEP), part-time care (12% more likely), center-based care (19% more likely),

and Head Start enrollment (2% more likely). Children with disabilities are less likely to be in

only parental care, slightly less likely to be in informal care, and more likely to be in center-

based care than typically developing peers. The magnitude of the estimates is moderate for

children with a continuing disability and rather large for children with an IEP. Children with a

disability on average pay less for child care than their peers; the hourly cost estimates indicate

that the lower cost of care is not simply an artifact of children with disabilities being in care

fewer hours each week.12

12 In order to focus discussion on the disability measures, and because the covariate estimates are consistent and expected across models, we do not present covariate estimates here. Appendix C contains covariate estimates for the model using the indicator for disability in any wave, excluding the state fixed effects estimates. In general, these are in the direction we would expect. Additional covariate estimates available upon request.


26

Table 4: Regression Models Predicting Child Care Outcomes at Wave 3

Regular

Care

Hours in

Care

FT

PT

Center

Head Start

Hom

e-B

ased

Cost/H

our

Annual

Cost

Any Wave 0.04*** 0.05 –0.01 0.04*** 0.06*** 0.01** 0.00 –0.41*** –386.75***

(0.01) (0.53) (0.01) (0.01) (0.01) (0.01) (0.01) (0.08) (87.54) Continuing Disability 0.09*** 0.99 –0.00 0.08*** 0.11*** 0.01 –0.01 –0.65*** –649.28***

(0.02) (0.79) (0.02) (0.02) (0.02) (0.01) (0.01) (0.10) (105.75)

IEP Indicator 0.18*** 2.32** 0.02 0.12*** 0.19*** 0.02*** –0.02 –1.10*** –

1,055.15***

(0.03) (0.93) (0.02) (0.02) (0.02) (0.01) (0.01) (0.11) (119.72)

Mean [0.68] [23.90] [0.43] [0.25] [0.50] [0.04] [0.05] [2.21] [2,565.34]

Multinomial Logit Predicting

Care Type Marginal Effect Estimate for

Care Type

Hom

e-B

ased C

are vs. N

o Care

Center-

Based vs.

No C

are

Hom

e-B

ased vs. C

enter

No C

are

Hom

e

Center-

Based

Any Wave 1.01 1.29*** 0.78*** –

0.04*** –0.02* 0.06***

(0.09) (0.09) (0.06) (0.01) (0.01) (0.01) Continuing Disability 1.19 1.77*** 0.67***

–0.09*** –0.02 0.12***

(0.16) (0.19) (0.08) (0.02) (0.02) (0.02)

IEP Indicator 1.64*** 2.85*** 0.58*** –

0.18*** –0.02 0.19***

(0.29) (0.40) (0.08) (0.03) (0.02) (0.03) *** p<0.01, ** p<0.05, * p<0.1; Standard errors in parentheses. Regular care indicates a child was in nonparental care at least ten hours each week; PT indicates part-time care (between 10-29 hours/week); FT indicates full-time nonparental care (30 hours/week or more). Models were all analyzed separately. Results are presented as marginal effects for binary outcomes or regression coefficients for continuous outcomes. Multinomial logistic models presented as odds ratios and overall marginal effects, respectively. Models predicting cost have sample sizes of 5,350; all others use sample of 6,600 (due to missing values for covariates). Dependent variable means are in brackets and unweighted. Source: Authors’ calculations using ECLS-B data In general, when models include the subsidy indicator, the magnitude of the estimates decreases

slightly (Table 5). Subsidy receipt decreases the likelihood of full-time care use and increases the

likelihood of part-time care for children with disabilities in any wave. Subsidies are associated

with decreased center-based care for children with continuing disabilities and increased use of

home-based care.


27

Table 5: Regression Models Predicting Child Care Outcomes at Wave 3, Subsidy Models

Regular C

are

Hours in C

are

FT

PT

Center

Head Start

Hom

e-Based

Cost/H

our

Annual C

ost

Any Wave 0.00 –0.63 –0.01 0.01 0.02 0.00 0.00 –0.25** –206.37*

(0.02) 0.62 (0.02) (0.01) (0.02) (0.01) (0.01) (0.11) (115.59) Marginal effect of subsidy 0.02 –0.91 –0.07** 0.09*** –0.01 0.01 0.00 0.05 –59.05 Continuing Disability 0.04 –0.4 –0.01 0.05** 0.09*** 0.00 –0.02** –0.44** –459.60***

(0.02) –0.99 (0.02) (0.02) (0.03) (0.01) (0.01) (0.18) (168.93) Marginal effect of subsidy –0.01 –0.64 –0.07* 0.06 –0.10** –0.01 0.04** 0.15 164.63

*** p<0.01, ** p<0.05, * p<0.1; Standard errors in parentheses. Regular care indicates a child was in nonparental care at least ten hours each week; PT indicates part-time care (between 10-29 hours/week); FT indicates full-time nonparental care (30 hours/week or more). Models were all analyzed separately. Results are presented as marginal effects for binary outcomes or regression coefficients for continuous outcomes. Models predicting cost have sample sizes of 5,350; all others use sample of 6,600. Source: Authors’ calculations using ECLS-B data

We also examine difference in child care use by disability type (Table 6, top panel). When

results are broken out by disability by type, it is clear that the overall disability estimates obscure

some heterogeneity by disability type; estimates of the effects of different diagnoses differ in

magnitude and, in some cases, direction. Broadly, the different diagnostic groups are related to

increased use of care overall, increased likelihood of part-time care, increased use of center-

based care, and decreased annual and hourly costs. However, some diagnostic groups, including

autism and physical impairments, are not significantly associated with use of regular nonparental

care, though for some, like congenital syndromes, the magnitude of the relationship is quite Table 6: Regression Models Predicting Child Care Outcomes at Wave 3, Subgroups

Any C

are

Hours in C

are

FT

PT

Center

Head

Start

Hom

e-B

ased

Cost/

Hour

Annual

Cost

Panel 1: Full Sample by Disability Type Communication Group 0.09*** 1.09* 0.00 0.08*** 0.10*** 0.01 0.00 –0.65*** –621.95***

(0.02) (0.64) (0.02) (0.01) (0.02) (0.01) (0.01) (0.09) (99.07) Intellectual Disability 0.14* 2.10 –0.02 0.12** 0.19*** 0.02 –0.05 –1.06*** –997.27***

(0.07) (3.76) (0.06) (0.05) (0.07) (0.03) (0.05) (0.21) (342.99) Emotional/Mental Health 0.11*** 1.10 –0.00 0.10*** 0.14*** 0.01 –0.01 –0.78*** –769.43***


28

(0.02) (0.86) (0.02) (0.02) (0.02) (0.01) (0.01) (0.12) (121.67)

Autism 0.09 0.08 –0.05 0.11*** 0.13** 0.03 –0.03 –1.20*** -1,631.23***

(0.06) (1.98) (0.05) (0.04) (0.06) (0.03) (0.03) (0.32) (274.87)

Chronic Condition 0.04* 1.25 0.03 0.01 0.04* –0.00 0.01 –0.22* 10.62

(0.02) (0.96) (0.02) (0.02) (0.02) (0.01) (0.01) (0.12) (157.87) Physical Impairment –0.05 –1.19 –0.05** 0.05** 0.05* 0.04*** –0.02 –0.52*** –465.12***

(0.11) (1.04) (0.03) (0.02) (0.03) (0.01) (0.01) (0.13) (156.76) Congenital Syndromes 0.18*** –8.06** –0.21 0.12 0.05 0.02 — –1.14*** –558.57

(0.03) (3.81) (0.15) (0.09) (0.12) (0.06) (0.42) (1,011.36)

Mean [0.68] [23.90] [0.43] [0.25] [0.50] [0.04] [0.05] [2.21] [2,565.34]

Panel 2: HH Income Ever < Poverty

Any Wave 0.01 –1.12 –0.03 0.04** 0.04* 0.02 0.00 –0.17** –200.29**

(0.02) (0.92) (0.02) (0.02) (0.02) (0.01) (0.01) (0.07) (84.63) Continuing Disability 0.07** –0.53 –0.02 0.08*** 0.10*** 0.01 0.01 –0.35*** –472.45***

(0.03) (1.24) (0.03) (0.02) (0.03) (0.02) (0.01) (0.10) (92.47)

IEP Indicator 0.26*** 2.78** 0.05 0.13*** 0.24*** 0.04** 0.01 –0.42*** –543.27***

(0.05) (1.34) (0.04) (0.03) (0.04) (0.02) (0.02) (0.09) (109.26)

Mean [0.66] [23.91] [0.43] [0.23] [0.49] [0.08] [0.03] [0.68] [887.99]

Panel 3: HH Ever Received SSI/SSDI

Any Wave 0.11*** 2.37 0.04 0.07* 0.17*** 0.04* 0.01 –0.59** –363.14**

(0.04) (1.75) (0.04) (0.03) (0.04) (0.02) (0.02) (0.23) (171.42) Continuing Disability 0.13*** 2.32 0.02 0.10*** 0.20*** 0.01 –0.02 –0.75*** –861.03***

(0.04) (2.01) (0.04) (0.04) (0.04) (0.03) (0.03) (0.19) (226.91)

IEP Indicator 0.26*** 5.67*** 0.08* 0.13*** 0.28*** 0.04 –0.01 –0.76*** –978.78***

(0.06) (1.96) (0.04) (0.04) (0.05) (0.03) (0.03) (0.19) (250.39)

Mean [0.68] [24.37 [0.44] [0.25] [0.52] [0.07] [0.03] [0.83] [1,001.53] *** p<0.01, ** p<0.05, * p<0.1. Standard errors in parentheses. Regular care indicates a child was in nonparental care at least ten hours each week; PT indicates part-time care (between 10-29 hours/week); FT indicates full-time nonparental care (30 hours/week or more). Results are presented as marginal effects for binary outcomes or regression coefficients for continuous outcomes. In the top panel, models predicting cost have sample sizes of 5,350; all other models in the top panel use sample of 6,900. In the middle panel, models predicting cost have sample sizes of 1,875; all other models in the middle panel have sample sizes of 2,550. In the bottom panel, models predicting cost have sample sizes of 550; all other models in the middle panel have sample sizes of 750. Dependent variable means are in brackets and unweighted. Source: Authors’ calculations using ECLS-B data large. While communication group conditions are associated with a slight increase in the number

of weekly hours in care, congenital syndromes are associated with a moderate decrease. Physical

impairments are the only diagnostic group significantly associated with a decreased use of full-

time care and a significantly increased likelihood of Head Start enrollment. Notable differences


29

are also evident in costs by diagnostic type, but all groups are associated with lower cost, with

the discount ranging from $465 annually (physical impairments) to $1600 annually (autism).

To better understand child care use among the potential SSI population, we repeat these

estimates for the subgroup of families who have ever had income below the poverty threshold in

any of the first three waves of the survey and families who report ever having received disability

benefits (Table 6, bottom panels). The pattern of results remains similar for both subgroups,

though the magnitude of the relationships is generally larger for both than it is for the population

overall, particularly for children who have an IEP. The estimates for likelihood of center-based

care for children in families who received disability benefits are notably large.

We graphed the unadjusted mean maternal employment rates for children with disabilities and

typically developing children, using the two main disability measures by survey wave, to better

understand patterns over time (Figures 1a, 1b). These figures suggest that as children age, rates

of maternal employment increase, which follows previous literature and our understanding of the

availability of child care for all children. Mothers of children with disabilities have lower rates of

overall employment than mothers of typically developing children. These figures also provide

some indication that the parallel trends assumption required for unbiased DD estimates is met for

these data; the two groups follow relatively similar patterns across survey waves. They do

indicate, however, that the biggest change in the rates of employment for mothers of children


30

Figures 1a, 1b: Maternal Employment Rates Across Survey Waves by Disability Status

Source: Authors’ calculations using the ECLS-B

0%10%20%30%40%50%60%70%80%90%

100%

9 Months (wave 1) 2 years (wave 2) 4 years (wave 3) Kindergarten (wave 4/5)Prop

ortio

n of

Mot

hers

Who

are

Em

ploy

ed

Typically-Developing Continuing Disability

0%10%20%30%40%50%60%70%80%90%

100%

9 Months (wave 1) 2 years (wave 2) 4 years (wave 3) Kindergarten (wave 4/5)Prop

ortio

n of

Mot

hers

Who

Are

Em

ploy

ed

Typically-Developing Any Disability


31

Table 7: Unadjusted Maternal Employment Rates by Wave, Weighted

Employed

FT Employment

PT Employment

Disability, Any Wave

Typically Developing Difference





Wave 2 53.9% 55.4% 1.5% 33.2% 35.6% 2.4% 20.7% 19.8% –0.8%

Wave 3 58.6% 60.1% 1.4% 40.1% 40.7% 0.6% 18.6% 19.4% 0.8%

Change 4.7% 4.7% 0.1% 6.9% 5.1% 1.8% –2.1% –0.4% 2.2%

Wave 3 58.6% 60.1% 1.4% 40.1% 40.7% 0.6% 18.6% 19.4% 0.8%

Kindergarten 60.2% 63.9% 3.7% 41.7% 43.8% 2.1% 18.5% 20.1% 1.6%

Change 1.5% 3.8% –2.3% 1.6% 3.1% –1.5% –0.1% 0.7% –0.7% FT Employment indicates household mother was employed at least 35 hours/week. PT Employment indicates household mother was employed less than 35 hours/week. Source: Authors’ calculations using the ECLS-B


32

with disabilities occurs between wave 2 (when the child is two years old) and the preschool

wave, rather than in kindergarten, as hypothesized.

At kindergarten enrollment, mothers of children with disabilities increase their employment at a

lower rate than mothers of typically developing children (Table 7). Beginning in wave 2, there is

a 1.5 percentage point difference in employment levels between the two groups overall. This

finding disguises the heterogeneity in the differences between full- and part-time employment

between the two groups; there is a larger gap in full-time employment (2 percentage points) and

a larger proportion of mothers of children with disabilities have part-time employment compared

to mothers of typically developing children. Between wave 2 and wave 3, both groups increased

employment overall at the same rate, though mothers of children with disabilities increased full-

time employment and decreased part-time employment at a higher rate than mothers of typically

developing children (seven percentage points in full-time employment compared to five

percentage points; two percentage point decrease in part-time employment compared to .4

percentage point). Between wave 3 and kindergarten enrollment, however, mothers of typically

developing children increased employment at a slightly higher rate (3.8 compared to 1.5

percentage points), almost all of the growth in full-time employment. There is essentially no

change in part-time employment for mothers of children with disabilities between preschool and

kindergarten enrollment, and a .7 percentage point increase for mothers of typically developing

children.

To further examine the relationship between disability and maternal employment at the two

different waves, Appendix D contains the results of logistic regressions predicting maternal

employment in the preschool wave (wave 3) and at kindergarten enrollment by the two main

disability measures. Generally, the marginal effects estimates of disability status for any maternal

employment for both measures in the preschool wave are nonsignificant. At the kindergarten

wave, however, both measures are significantly associated with a three percentage point decrease

in the likelihood of any maternal employment. Both are also associated with a decrease in the

average maternal hours worked and average total parental hours worked. Results are similar for


33

Table 8: DD Analysis using LPM predicting maternal employment

Any Employment FT Employment PT Employment Hours Total Hours

Any

Continuing

IEP

Any

Continuing

IEP

Any

Continuing

IEP

Any

Continuing

IEP

Any

Continuing

IEP

Panel 1: Full Sample

Disability

** –0.0261

–0.0311 –0.0141

** –0.0320

*** –0.0467

* –0.0302

0.0063 0.0023 0.0189

*** –1.5104

** –1.6092

–1.1633

*** –2.4557

** –2.0491

*** –2.7377

(0.0128) (0.0192) (0.0189) (0.0131) (0.0161) (0.0178) (0.0109) (0.0158) (0.0169) (0.5233) (0.7753) (0.7097) (0.6418) (0.9318) (0.9506)

Kindergarten

*** 0.0214

*** 0.0188

*** 0.0188

0.0116

* 0.0123

0.0105 0.0102 0.0069 0.0088

*** 0.7871

*** 0.8123

*** 0.7736

*

0.6848

** 0.8291

** 0.7311

(0.0071) (0.0066) (0.0066) (0.0071) (0.0066) (0.0066) (0.0070) (0.0066) (0.0065) (0.2771) (0.2570) (0.2551) (0.3803) (0.3518) (0.3467)

DD

–0.0120 –0.0073 –0.0117 0.0022 0.0116 0.0186 –0.0143 –0.0059 –0.0317 –0.0208 –0.3159 0.0258 0.1784 –0.9658 –0.0937

(0.0120) (0.0179) (0.0212) (0.0118) (0.0167) (0.0226) (0.0119) (0.0166) (0.0208) (0.4625) (0.6957) (0.8361) (0.6266) (0.9414) (1.1795)

Panel 2: HH Income Ever < Poverty

DD

0.0034 –0.0036 –0.0291 0.0197 0.0095 –0.0057 –0.0174 –0.0109 –0.0267 0.4491 –0.2671 –0.2612

0.3202 –1.6562 –0.2706

(0.0224) (0.0310) (0.0380) (0.0223) (0.0313) (0.0394) (0.0199) (0.0273) (0.0334) (0.8836) (1.1985) (1.5164)

(1.1784) (1.6563) (2.1812) Panel 3: HH Ever Received SSI/SSDI

DD

0.0078 –0.0316 –0.0556 0.0289 –0.0325 –0.0343 –0.0238 0.0007 –0.0232 0.9649 –1.5668 –1.9047 0.6690 –2.7590 –1.6716

(0.0378) (0.0407) (0.0429) (0.0374) (0.0423) (0.0465) (0.0333) (0.0385) (0.0424) (1.5100) (1.5964) (1.7132)

(2.0564) (2.2617) (2.5534) *** p<0.01, ** p<0.05, * p<0.1 Standard errors in parentheses. FT Employment indicates household mother was employed at least 35 hours/week; PT Employment indicates household mother was employed less than 35 hours/ week. Hours is a measure of regular weekly hours worked by household mother; Total Hours is a measure of combined weekly hours for all parents in the households. The full sample has 6,700 observations. The sub-sample of households ever below poverty includes 2,550 observations. The sub-sample of households that have ever received disability benefits has 750 observations. Source: Authors’ calculations using ECLS-B


34

children who live in households that have ever had income below poverty and those that have

ever received SSI or SSDI (second and third panels).

We examined the differences in the rates of change of maternal employment at kindergarten

enrollment for children with and without disabilities using DD estimates for linear probability

models (LPM) that predict any maternal employment, full-time employment, and part-time

employment, as well as models predicting maternal and total parental hours worked in the last

week (Table 8). The estimates for the main effects of having a disability are negative in all

models, except for the LPM predicting part-time employment, and mostly statistically

significant. Childhood disability, depending on the measure, is associated with an approximately

three to four percentage point decrease in maternal employment overall and in full-time

`employment at kindergarten, as well as one-to-two fewer average hours worked each week for

both mothers and all parents in the household. The estimates for the effect of kindergarten

enrollment on employment outcomes are generally positive and statistically significant. The DD

estimator estimates the difference in the change of employment rates for mothers of children

with disabilities compared to mothers of typically developing children. For the most part, the DD

estimates are not statistically significant; all are close to zero. The direction of the estimates is

negative for overall employment and part-time employment. This finding suggests that there is

no statistically significant difference in the change in the rate of maternal employment at

kindergarten enrollment for mothers of children with disabilities compared to mothers of

typically developing children. We find a similar pattern for poor households and households that

have ever received disability benefits (second and third panels).

4.2. Alternate Specifications

We tested our DD models using a range of alternative specifications. First, we included wave 2

of the data in our estimates; this specification results in a statistically significant negative

estimate for the rate of change in part-time maternal employment for mothers of children with

disabilities compared to other mothers. This finding is consistent when we apply the

recommended survey weights whether or not we include wave 2. We find similar results when

we run models excluding children in half-day kindergarten; in this specification, mothers of


35

children with IEPs increase their rates of part-time employment at kindergarten by approximately

four percentage points less than mothers of typically-developing children.

We also estimated models using imputed data for the main analytic sample and imputed data

across waves for all 10,700 of the original participants. Using the imputed data, results from

some models suggest a statistically significant but smaller rate of change in maternal

employment for mothers of children with disabilities. Using imputed data, we estimate that

mothers of children with a disability in any wave increased their rate of overall employment less

than mothers of typically developing children by approximately two percentage points, with a

similar estimate for part-time employment. The imputed data also suggest that the rate by which

mothers of children with continuing disabilities increased their rate of full-time employment was

four percentage points lower than it was for mothers of typically developing children. Contrary

to our initial hypothesis, these findings suggest employment at kindergarten for mothers of

typically developing children increases at a higher rate than for mothers of children with a

disability diagnosis in any wave prior to kindergarten.

Because of the increase in child care use at the preschool wave for all children, but for children

with disabilities in particular, we estimated models using the preschool year as the treatment year

to determine whether availability of care during this developmental stage may differentially

impact maternal employment. This analysis did not yield any significant results. Due to

endogeneity concerns related to early identification and use of care (i.e., children in care may

also be more likely to be identified as having a disability), we also tested a model where we

limited children with a disability only to those whose condition was identified prior to entering

child care; this analysis does not find any differences in results.

We also examined differences by diagnostic categories for the LPMs predicting differences in

maternal employment rates at kindergarten enrollment (Table 9). Looking at main effects, all of

the categories are associated with a decreased likelihood of maternal employment but with


36

differences in the relationship to maternal employment by childhood conditions.13 Though the

DD estimates for all conditions are nonsignificant, the magnitude and, in some cases, the

direction differs notably from the overall disability estimates. Indeed, when we use imputed data,

we find a positive and statistically significant estimate for employment for mothers of children

with autism.

Table 9: DD Analysis using LPM predicting maternal employment by disability category

Main Effect DD Estimator

Communication –0.0423*** –0.0028

(0.0160) (0.0148)

IDD –0.1572** 0.0441

(0.0672) (0.0572)

Emotional/Mental Health –0.0414* 0.0014

(0.0231) (0.0200)

Autism –0.0382 0.0443

(0.0577) (0.0368)

Chronic –0.0134 –0.0126

(0.0216) (0.0211)

Physical/Orthopedic –0.0465* –0.0144

(0.0263) (0.0244)

Congenital Syndromes –0.2259** 0.2018

(0.1113) (0.1342) *** p<0.01, ** p<0.05, * p<0.1 Standard errors in parentheses Source: Authors’ calculation using the ECLS-B

5. Discussion At age four, children with disabilities are more likely to be in any regular nonparental care, more

likely to be in part-time care, more likely to be enrolled in center-based care as their primary

arrangement, more likely to attend Head Start, and more likely to have arrangements that cost

less on an hourly and annual basis compared to typically developing children. Children with

disabilities are no more or less likely to be enrolled in full-time care than their typically

developing peers; there is some indication they are less likely to be enrolled in informal care.

13 We do not show the main effects of kindergarten enrollment because these estimates do not differ substantially across models, nor do they differ substantially from the main models.


37

Notably, the pattern of results is broadly similar no matter which definition of disability is

employed, with the exception of the outcome of Head Start enrollment. As expected, our two

non-IEP measures for disability bound estimates; the measure for disability in any wave

generally results in estimates of lower magnitude than those for continuing disability. These

findings provide some confidence in our selection of disability measures. Receipt of an IEP is

associated with the largest magnitude estimates. Results are robust when we control for

employment and across measures of income.

We exclude the subsidy indicator from our main models due to concerns about endogeneity (e.g.,

in order to receive a subsidy, children must be enrolled in care). However, when we model

subsidy receipt, the magnitude of the relationship between disability and child care use is

decreased moderately, particularly for the disability in any wave measure (our lower-bound

measure), though the pattern of relationships remains the same. Given the disproportionate

number of children with disabilities who report subsidized care—approximately 10 to 20

percentage points higher than typically developing children—this is to be expected. To further

probe the effect of subsidies, we also examined the marginal effect of subsidies by disability

status. We find that subsidy receipt is associated with a decreased likelihood of full-time child

care for both measures of disability. For children with a continuing disability, subsidy receipt is

associated with a decreased likelihood of center-based care and an increased likelihood of home-

based care. The negative relationship with center-based care is surprising; we might expect

subsidies to increase access to center-based care.

The patterns of care access and use do not differ when we consider only households that have

had income below the poverty threshold at any point in the child’s life—households that are

most likely to be meet the SSI eligibility criteria—and those that have ever received disability

benefits. In fact, the strength of the relationship between disability status and part-time care and

center-based care are stronger for these subgroups. This finding indicates that the relationship to

access is not being driven simply by high-income families choosing center-based care. This

analysis paints an encouraging picture of child care access for SSI recipients and the families


38

most likely to be eligible for SSI; these estimates suggest that the current child care landscape

may be sufficiently supporting employment for these families.

Our results are consistent with the conclusion that the constellation of child care policies may be

effectively supporting access to child care for children with disabilities. At the most basic level,

we see an increased likelihood of regular child care in general for children with disabilities. We

also find an increased likelihood of center-based care compared to other care types. In general,

much of the public policy infrastructure is designed to improve access to center-based care; thus,

this relationship may be a result of policy levers. Our results also suggest that children with

disabilities receive care at a decreased cost, all else constant, even compared to other families

with income below the poverty threshold. Though this may be related to the quality of care these

families can access, there is some indication that it is instead a result of policy. In particular, we

find that cost is decreased overall, both annually and hourly. We also see differences by

disability type; the cost of care for children with autism is discounted by the largest amount,

followed by children with intellectual disabilities and children with other congenital syndromes.

These conditions are often diagnosed prior to preschool, and we might expect to result children

with these conditions to have the most difficulty accessing affordable care. Thus, differences in

cost may be the result of effective public policy that provides access to early care and education

via IEP or subsidy.

Our estimates for explicitly policy-relevant measures are also consistent with policy increasing

access. Having a disability in any wave statistically significantly increases the likelihood of Head

Start enrollment, which is notable given the small number of children in our sample enrolled in

Head Start. The estimates for the IEP indicator are particularly telling. Having an IEP, which

indicates that the child is receiving services from the Department of Education under IDEA,

strongly increases the likelihood of care overall, center-based care, Head Start enrollment, and

decreased costs. These indicators are all consistent with increased access and in clear keeping

with the purpose of IDEA. The estimates for the effect of an IEP on care use for low-income

children and children in households that have received disability benefits are particularly

encouraging. Having an IEP increases the likelihood of any child care use and center-based care


39

by over 25 percentage points for these groups. Further, while receipt of an IEP is associated with

fewer hours in care for the overall population, it is associated with an increased number of hours

in care for these populations and an increased likelihood of full-time care for children in

households who have ever received disability benefits.

One issue to consider is the direction of the relationship between child care use and disability.

Children in child care—particularly high-quality, center-based care—are more likely to be

screened for health concerns; screening is required by law, under IDEA’s Child Find provision.

However, some of our estimates indicate that the relationship is not just about increased

identification for children in care. First, results are robust when we conducted analyses using

only children with diagnoses prior to care enrollment. Also, our results by diagnostic type are not

indicative of a story of increased identification once in care. Though the patterns are all broadly

similar to the overall results, children with intellectual disabilities and children with autism have

some of the strongest likelihoods of center-based care use. These conditions are generally

identified at an early age. The strength of these relationships is also stronger than the estimates

for communication-related conditions, which may be more likely to be identified and screened

for after a child has entered care. It is also possible that parents may select center-based care for

children with disabilities, even prior to diagnosis, because they think it will be helpful to their

children and as a means of connecting to services, including formal diagnoses and IEP receipt.

Although, our approach cannot rule out that parents with non-typically developing children

prefer center based care, and it is also possible that experiencing center-based care increases the

odds of a disability being formally diagnosed.

The association between disability status and use of part-time care rather than full-time care

warrants additional consideration. It may indicate that young children with disabilities are using

early childhood centers primarily as sites for services or learning enrichment rather than

traditional child care to support parental employment. In this case, policy may be succeeding in

supporting the child’s health needs, but it may not be providing care sufficient to support

parental employment. This pattern may also reflect the complexity families face in balancing

caregiving needs and other requirements like medical appointments or it may be a result of


40

children’s health conditions impacting their ability to participate in full-time care. Indeed, the

results of the DD analysis indicate that parents of children with disabilities are less likely to

increase their rates of part-time employment at kindergarten enrollment, perhaps because of their

overall higher rates of part-time employment in earlier waves. Finally, we may be concerned that

policy is effectively supporting part-time enrollment and access, but not full-time access. Indeed,

the marginal effect of subsidy receipt is associated with an increased likelihood of part-time

compared to full-time care. Children may have access to care for the portion of the day when

they receive IEP services, for example, but not for the rest of the day. This pattern could also be

the result of the expansion of public pre-K programs, which are typically not full day but do

lower the age of access to free, public education.

That we find a stronger association with part-time care use may also indicate that our findings

are more consistent with the previous literature emphasizing families’ challenges finding care

than it first appears. Though we find that children with disabilities are more likely to be enrolled

in care than children without disabilities, we have no insight into parents’ satisfaction with care,

the length of time the child has been in that particular child care arrangement (or how long they

will remain), nor the process of locating care and what public resources may be available to help

them in that process. Indeed, what evidence we do have, suggests that children with disabilities

enter care later than typically-developing peers, which may underscore the difficulty locating

care. The differences we find in care use by disability status also underscore that our findings

represent an average for all children identified with disabilities, which may understate the

difficulty some families have accessing care.

Overall, the results of our DD analysis could have several policy implications. These results,

taken in concert with the child care estimates, suggest child care is accessible for families raising

young children with disabilities. The constellation of federal child care policies may be

adequately supporting access to child care for families, particularly low-income families, raising

children with disabilities. Further, the increased availability of public pre-K programs for four-

year-olds may play a role. The unadjusted rates of employment indicate a greater shift in

employment between the two-year-old wave and the four-year-old wave, which may also suggest


41

this expansion has a role in the results. To test this, we also ran a model with preschool as the

treatment year; results were nonsignificant.

Additionally, the nonsignificant results and the overall trend of negative estimates may simply

indicate that the labor supply of parents of children with disabilities is less elastic than that of

parents of typically developing children. We know these parents have many more responsibilities

to juggle, and enrollment in kindergarten may not relieve these responsibilities. In fact, given the

length of school days and other out-of-school time for holidays and summer, kindergarten may

provide less support for employment for parents who have to find after-school or other out-of-

school care for children enrolled in school. It may also be important to consider other policy

levers to support employment for parents raising a child with a disability. Even if child care is

sufficiently available, parents’ ability to work may be hindered by the increased needs of

children with disabilities. Parents may not be able to take their children with disabilities to more

frequent medical appointments or address their overall needs if they do not have access to paid

leave or flexible scheduling. This limitation may be highlighted in the different patterns of

results for children with continuing disability in households that report ever having received

disability benefits; the DD estimates are larger and concentrated almost entirely in full-time

employment, meaning mothers of children with disabilities in this group are less likely to

increase employment at kindergarten enrollment compared to mothers of typically developing

children.

Finally, the DD estimates for specific diagnostic categories indicate the need to consider how

different conditions may differentially affect access to care and parental employment. Mothers of

children with autism appear to increase their employment at kindergarten enrollment at greater

rates than mothers of typically developing children. Though nonsignificant, the results suggest

this may also true for other conditions under the broad umbrella of developmental disability,

including intellectual disability and congenital syndromes. This finding may be a result of the

severity of the condition, which could influence care availability (though we do not see this in

the child care results), or it may be indicative of the role of diagnostic timing or differences in

how conditions manifest across developmental stages. All of these conditions can be identified


42

early, which may allow families time to account for services and supports by kindergarten

enrollment.

6. Limitations

The nature and severity of children’s special needs are clearly important to understanding child

care accommodations and parents’ employment contexts. Although the data offer a relatively

large sample of children with disabilities and we attempt to disaggregate by some diagnoses, we

are still unable to do so beyond broad categorizations. In addition, we rely on child care

arrangements and estimates of maternal employment rates to infer information about the

accessibility of child care. This approach does not account for parental preferences related to

child care; parents of children with disabilities may have different preferences for care than

children of typically developing peers. It also does not account for the parental employment

context, such as the availability of employer flexibility and paid leave, which may also influence

parental employment. Therefore, lower rates of maternal employment may simply reflect a

difference in parental preferences or the impact of related policies rather than accessibility of

care. The relationship between the child’s health status and parental employment characteristics

could also bias our estimates. Similarly, while this study aims to provide information about the

current menu of policies intended to assist families that have children with disabilities in

accessing child care, it cannot disentangle the roles of specific policies nor can it account for the

supply of child care. These data also do not reflect some recent, consequential changes in the

child care landscape, such as changes to the subsidies intended to increase access and

implementation of IDEA Part C, which extends services to infants and toddlers. Our data also

limit our understanding of some key timing issues; it may be useful to have data on post-

treatment outcomes, particularly since the kindergarten enrollment years coincide with the

beginning of the Great Recession. Additionally, though we have information for the first wave

that a parent reports their child having a specific health condition, we do not know precisely

when the child was diagnosed.

7. Summary and Conclusions


43

Our analysis broadly supports the idea that federal policies—including subsidies, IDEA, and

Head Start—are providing access to child care for children with disabilities, including SSI

recipients. Children with disabilities have higher likelihoods of enrollment in child care,

particularly part-time and center-based care, at age four, and there is some evidence that policy

levers may be easing access and decreasing costs. The results of our DD analysis are

nonsignificant and indicate, that, if anything, mothers of children with disabilities increase their

rate of employment at kindergarten enrollment at lower rates than mothers of typically

developing peers. Thus, our analysis finds support for the efficacy of policies designed to support

employment for parents of children with disabilities, including for children from very low-

income families, who may be most likely to receive SSI. Our findings also suggest several

avenues for future research, including examining child care access for very young children with

disabilities and the role of IDEA Part C in providing access; exploiting state differences in child

care policy to disentangle the role of specific policies; oversampling children with disabilities in

large-scale data collections to disaggregate by disability type; and exploring employment

supports available to parents of school-age children with disabilities, such as out-of-school care

and paid leave.

Children with disabilities, including SSI recipients and their families, appear to be using early

care and education at relatively high rates. This access may provide an important support for

parental employment and may also confer developmental benefits for children. As the number of

young children with disabilities continues to grow, it will be important for early care and

education providers to be appropriately trained to meet the needs of these children and their

families. It will also remain important for policymakers and federal agencies that serve children

with disabilities to monitor and continue to support child care access and provide other policy

supports to ensure families are receiving the support they need to maintain economic stability.


44

References Aizer, Anna, Nora E. Gordon, and Melissa Schettini Kearney. Exploring the Growth of the Child

SSI Caseload in the Context of the Broader Policy and Demographic Landscape. No.

odrc13-02. National Bureau of Economic Research, 2013.

Joffe-Walt, Chanah. “Trends with Benefits.” This American Life. Podcast audio, March 22,

2013.https://www.thisamericanlife.org/490/trends-with-benefits

Booth, Cathryn L., and Jean F. Kelly. "Child care and employment in relation to infants'

disabilities and risk factors." American Journal on Mental Retardation 104, no. 2 (1999):

117–130.

Booth-LaForce, Cathryn, and Jean F. Kelly. "Childcare patterns and issues for families of

preschool children with disabilities." Infants & Young Children 17, no. 1 (2004): 5–16.

Boyle, Coleen A., Sheree Boulet, Laura A. Schieve, Robin A. Cohen, Stephen J. Blumberg,

Marshalyn Yeargin-Allsopp, Susanna Visser, and Michael D. Kogan. "Trends in the

prevalence of developmental disabilities in US children, 1997–2008." Pediatrics 127, no.

6 (2011): 1034–1042.

Brandon, Peter D. "Child care utilization among working mothers raising children with

disabilities." Journal of Family and Economic Issues 21, no. 4 (2000): 343–364.

Brennan, Eileen M., and Ana Maria Brannan. "Participation in the paid labor force by caregivers

of children with emotional and behavioral disorders." Journal of Emotional and

Behavioral Disorders 13, no. 4 (2005): 237–246.

Ceglowski, Deborah Ann, Mary Ellin Logue, Annette Ullrich, and Jaesook Gilbert. "Parents’

perceptions of child care for children with disabilities." Early Childhood Education

Journal 36, no. 6 (2009): 497–504.


45

Child Care Aware of America. Parents and the High Cost of Child Care: 2017 Report.

Arlington, VA: Child Care Aware of America. (2017). https://usa.childcareaware.org/wp-

content/uploads/2017/12/2017_CCA_High_Cost_Report_FINAL.pdf

Costanzo, Molly A., and Katherine Magnuson. "How does disability influence child care

arrangements for young children? An examination using the NHES ECPP." Children and

Youth Services Review 99 (2019): 210–225.

DeRigne, LeaAnne, and Shirley Porterfield. "Employment change and the role of the medical

home for married and single-mother families with children with special health care

needs." Social science & medicine 70, no. 4 (2010): 631–641.

Deshpande, Manasi. "The effect of disability payments on household earnings and income:

Evidence from the SSI children's program." Review of Economics and Statistics 98, no. 4

(2016): 638–654.

DeVore, Simone, and Barbara Bowers. "Childcare for children with disabilities: Families search

for specialized care and cooperative childcare partnerships." Infants & Young Children

19, no. 3 (2006): 203–212.

Duggan, Mark G., and Melissa Schettini Kearney. "The impact of child SSI enrollment on

household outcomes." Journal of Policy Analysis and Management 26, no. 4 (2007):

861–886.

Forry, Nicole, Paula Daneri, and Grace Howarth. Child Care Subsidy Literature Review. OPRE

Brief 60. 2013.

Glenn-Applegate, Katherine, Jill Pentimonti, and Laura M. Justice. "Parents’ selection factors

when choosing preschool programs for their children with disabilities." Child & Youth

Care Forum 40, no. 3 (2011): 211–231.

Grisham-Brown, Jennifer, Megan Cox, Meg Gravil, and Kristen Missall. "Differences in child

care quality for children with and without disabilities." Early Education and

Development 21, no. 1 (2010): 21–37. http://doi.org/10.1080/10409280902783491


46

Guldi, Melanie, Amelia Hawkins, Jeffrey Hemmeter, and Lucie Schmidt. Supplemental Security

Income and Child Outcomes: Evidence from Birth Weight Eligibility Cutoffs. No.

w24913. National Bureau of Economic Research, 2018.

Halfon, Neal, Amy Houtrow, Kandyce Larson, and Paul W. Newacheck. "The changing

landscape of disability in childhood." The Future of Children (2012): 13–42.

Knoche, Lisa, Carla A. Peterson, Carolyn Pope Edwards, and Hyun-Joo Jeon. "Child care for

children with and without disabilities: The provider, observer, and parent perspectives."

Early Childhood Research Quarterly 21, no. 1 (2006): 93–109.

Lally, J. Ronald, Abbey Griffin, Emily Fenichel, Marilyn Segal, Eleanor Szanton, and Bernice

Weissbourd. Caring for Infants and Toddlers in Groups: Developmentally Appropriate

Practice. Washington, DC: Zero to Three, 1995.

Laughlin, Lynda. Who's Minding the Kids? Child Care Arrangements, Spring 2011. US

Department of Commerce, Bureau of the Census, 2013.

Lemmon, Megan Elaine. "How Young Children's Disabilities Affect Parents' Labor Force

Participation and Earnings." Masters’ Thesis, Pennsylvania State University, 2015.

Lindley, Lisa C., and Barbara A. Mark. "Children with special health care needs: Impact of

health care expenditures on family financial burden." Journal of Child and Family

Studies 19, no. 1 (2010): 79–89.

Loprest, Pamela, and Amy Davidoff. "How children with special health care needs affect the

employment decisions of low-income parents." Maternal and Child Health Journal 8, no.

3 (2004): 171–182.

Lukemeyer, Anna, Marcia K. Meyers, and Timothy Smeeding. "Expensive children in poor

families: Out‐of‐pocket expenditures for the care of disabled and chronically ill children

in welfare families." Journal of Marriage and Family 62, no. 2 (2000): 399–415.


47

Mitra, Sophie, Michael Palmer, Hoolda Kim, Daniel Mont, and Nora Groce. "Extra costs of

living with a disability: A review and agenda for research." Disability and Health Journal

10, no. 4 (2017): 475–484.

National Institute for Early Education Research. State of Preschool 2018: State Preschool

Yearbook. Rutgers Graduate School of Education, 2019. http://nieer.org/wp-

content/uploads/2019/04/YB2018_Executive-Summary.pdf

Odom, Samuel L., Virginia Buysse, and Elena Soukakou. "Inclusion for young children with

disabilities: A quarter century of research perspectives." Journal of Early Intervention 33,

no. 4 (2011): 344–356.

Office of the Assistant Secretary of Planning. Department of Health and Human Services.

National Trends in the Child SSI Program. ASPE Research Brief. 2015.

Office of Head Start. Poverty Guidelines and Determining Eligibility for Participation in Head

Start Programs. Head Start Early Childhood Learning and Knowledge Center, 2019.

https://eclkc.ohs.acf.hhs.gov/eligibility-ersea/article/poverty-guidelines-determining-

eligibility-participation-head-start

Parish, Susan L., and Jennifer M. Cloud. "Financial well-being of young children with

disabilities and their families." Social Work 51, no. 3 (2006): 223–232.

Parish, Susan L., Jennifer M. Cloud, Jungwon Huh, and Ashley N. Henning. "Child care,

disability, and family structure: Use and quality in a population-based sample of low-

income preschool children." Children and Youth Services Review 27, no. 8 (2005): 905–

919.

Parish, Susan L., Roderick A. Rose, Michal Grinstein-Weiss, Erica L. Richman, and Megan E.

Andrews. "Material hardship in US families raising children with disabilities."

Exceptional Children 75, no. 1 (2008): 71–92.

Porterfield, Shirley L. "Work choices of mothers in families with children with disabilities."

Journal of Marriage and Family 64, no. 4 (2002): 972–981.


48

Powers, Elizabeth T. "New estimates of the impact of child disability on maternal employment."

American Economic Review 91, no. 2 (2001): 135–139.

Powers, Elizabeth T. "Children’s health and maternal work activity estimates under alternative

disability definitions." Journal of Human Resources 38, no. 3 (2003): 522–556.

Romig, K. SSI: A lifeline for children with disabilities. Center on Budget and Policy Priorities,

May 11, 2017. https://www.cbpp.org/research/social-security/ssi-a-lifeline-for-children-

with-disabilities

Shattuck, Paul T., and Susan L. Parish. "Financial burden in families of children with special

health care needs: Variability among states." Pediatrics 122, no. 1 (2008): 13–18.

Snow, Kyle, Azucena Derecho, Sara Wheeless, J. Lennon, J. Rosen, J. Rogers, S. Kinsey, K.

Morgan, and P. Einaudi. "Early Childhood Longitudinal Study, Birth Cohort (ECLS-B),

Kindergarten 2006 and 2007 data file user’s manual (2010-010)." Washington, DC:

National Center for Education Statistics, Institute of Education Sciences, US Department

of Education, 2009.

Social Security Administration. SSI Annual Statistical Report, 2017. Table 19. Released

September 2018. https://www.ssa.gov/policy/docs/statcomps/ssi_asr/2017/

Social Security Administration. Understanding Supplemental Security Income SSI for Children—

2019 Edition. Understanding SSI, 2019. https://www.ssa.gov/ssi/text-child-ussi.htm

Solon, Gary, Steven J. Haider, and Jeffrey M. Wooldridge. "What are we weighting for?"

Journal of Human Resources 50, no. 2 (2015): 301–316.

Stabile, Mark, and Sara Allin. "The economic costs of childhood disability." The Future of

Children 22, no. 1 (2012): 65–96.

Sullivan, Amanda L., Elyse M. Farnsworth, and Amy Susman-Stillman. "Patterns and predictors

of childcare subsidies for children with and without special needs." Children and Youth

Services Review 88 (2018): 218–228.


49

US Department of Education. Programs: Early Intervention Program for Infants and Toddlers

with Disabilities. Office of Special Education Programs, 2016.

https://www2.ed.gov/programs/osepeip/index.html?exp=0

US Department of Health and Human Services. Child Care and Development Fund Report to

Congress for FY2002-FY2003. Child Care Bureau, 2003.

US Department of Health and Human Services. CCDF Reauthorization Frequently Asked

Questions—Archived. Office of Child Care, 2015.

https://www.acf.hhs.gov/occ/resource/ccdf-reauthorization-faq-archived

US Department of Health and Human Services and US Department of Education. Policy

Statement on Inclusion of Children with Disabilities in Early Childhood Programs.

Statement, 2015. https://www2.ed.gov/policy/speced/guid/earlylearning/joint-statement-

full-text.pdf

U.S. Library of Congress. Congressional Research Service. The Individuals with Disabilities

Education Act (IDEA) Funding: A Primer, by Kyrie E. Dragoo. CRS Report R44624.

2016.

Wall, Shavaun, Ellen E. Kisker, Carla A. Peterson, Judith J. Carta, and Hyun-Joo Jeon. "Child

care for low-income children with disabilities: Access, quality, and parental satisfaction."

Journal of Early Intervention 28, no. 4 (2006): 283–298.

Wasi, Nada, Bernard van den Berg, and Thomas C. Buchmueller. "Heterogeneous effects of

child disability on maternal labor supply: Evidence from the 2000 US Census." Labour

Economics 19, no. 1 (2012): 139–154.

Weglarz-Ward, Jenna M., and Rosa Milagros Santos. "Parent and Professional Perceptions of

Inclusion in Childcare." Infants & Young Children 31, no. 2 (2018): 128–143.

Weglarz-Ward, Jenna M., Rosa Milagros Santos, and Jennifer Timmer. "Factors that support and

hinder including infants with disabilities in child care." Early Childhood Education

Journal 47, no. 2 (2019): 163–173.


50

Appendix A: Conditions By Interview Wave

Condition Waves Included Asthma All Blindness 1,2 Difficulty seeing All Difficulty hearing All Cleft palate 1 Heart defect All Failure to thrive 1 Problem with mobility All Problem using hands 1 Down Syndrome 1 Turner Syndrome 1 Spina bifida 1 Other special need 1 Crossed eyes 2 Delay in walking 2 Delay in talking 2 Other delay 2–5 Epilepsy 2–5 Intellectual disability 2–5 Requires special equipment 2–5 Condition impairs play 2 Evaluated and diagnosed problem with attention 3–5 Evaluated and diagnosed problem with activity level 3–5 Evaluated and diagnosed problem using limbs 3–5 Evaluated and diagnosed problem with communication 3–5 Autism 3–5 Oppositional Defiant Disorder 3–5 ADHD 3–5 Blood disease 4–5 Other chronic condition 4–5

Source: ECLS-B


51

Appendix B: Models Predicting Child Care Arrangements in Waves 1 & 2

Informal and Home-Based Care vs No Care

Center-Based vs No Care

Informal and Home-Based vs Center No Care Informal/Home Center-Based

Wave 1 0.89 0.95 0.94 0.02 –0.02 –0.00 (0.07) (0.13) (0.13) (0.02) (0.02) (0.01)

Wave 2 0.99 1.17 0.85 –0.01 –0.01 0.02 (0.09) (0.13) (0.10) (0.02) (0.02) (0.01)

*** p<0.01, ** p<0.05, * p<0.1 Standard errors in parentheses. Regular care indicates a child was in nonparental care at least ten hours each week; PT indicates part-time care (between 10-29 hours/week); FT indicates full-time nonparental care (30 hours/week or more). Models use wave specific disability measures and outcomes. Source: Authors’ calculations using the ECLS-B

Models Predicting Child Care Arrangements in Waves 1 & 2

Regular Carea Hours in Careb FT a PT a Center a Home-Based a Annual Costb

Wave 1 –0.02 –0.60 –0.02 0.00 0.00 –0.01 119.17

(0.02) (0.86) (0.02) (0.01) (0.01) (0.01) (243.56)

Wave 2 0.01 0.53 0.00 0.01 0.02 0.02 –106.68

(0.02) (0.71) (0.02) (0.01) (0.01) (0.01) (175.83)


52

Appendix C: Covariate Estimate for Models Predicting Child Care Use

in Wave 3

Covariate Estimates for Models Predicting Child Care Use in Wave 3

Regular Carea

Hours in Careb FT a PT a Centera

Head Start a

Home-Baseda

Annual Costb

Any Disability, Before Kindergarten 0.04*** 0.05 –0.01 0.04*** 0.06*** 0.01** 0.00 –386.75***

(0.01) (0.53) (0.01) (0.01) (0.01) (0.01) (0.01) (87.54)

Male 1.13** 0.31 1.06 1.06 1.07 1.02 1.04 –118.70

(0.06) (0.47) (0.06) (0.06) (0.06) (0.13) (0.13) (78.22)

Child Race/Ethnicity, White reference reference reference reference reference reference reference reference

Black 1.73*** 8.08*** 2.55*** 0.51*** 1.78*** 3.31*** 0.96 –203.41

(0.17) (0.81) (0.24) (0.06) (0.16) (0.67) (0.23) (127.86)

Latinx 1.09 1.01 1.22** 0.86* 1.06 1.61** 0.91 –186.79

(0.10) (0.74) (0.10) (0.08) (0.09) (0.34) (0.17) (132.93)

Other Race/Ethnicity 1.34*** 3.53*** 1.46*** 0.86* 1.39*** 1.43* 0.64** 471.86***

(0.11) (0.66) (0.11) (0.07) (0.10) (0.30) (0.11) (122.94)

Single Parent Family 1.92*** 8.66*** 2.30*** 0.70*** 1.12 1.25 1.58*** 522.63***

(0.15) (0.66) (0.17) (0.06) (0.08) (0.20) (0.27) (104.41)

Number of Household Children 5 and Under 0.79*** –2.29*** 0.78*** 1.01 0.90*** 1.14* 0.67*** –296.85***

(0.03) (0.32) (0.03) (0.04) (0.03) (0.09) (0.06) (51.26)

Number of Household Children 6–17 0.87*** –1.55*** 0.88*** 0.99 0.89*** 1.00 0.99 –171.63***

(0.02) (0.24) (0.03) (0.03) (0.02) (0.06) (0.06) (35.84)

Number of Nonparental Adults in Household 1.11*** 2.08*** 1.18*** 0.90*** 0.94* 1.19*** 0.73*** –109.45**

(0.04) (0.30) (0.04) (0.04) (0.03) (0.08) (0.07) (46.20)

Household Income <100%FPL 0.87 –3.74*** 0.67*** 1.42*** 1.30*** 1.75** 0.37*** –1,316.52***


53

Covariate Estimates for Models Predicting Child Care Use in Wave 3

Regular Carea

Hours in Careb FT a PT a Centera

Head Start a

Home-Baseda

Annual Costb

(0.08) (0.79) (0.06) (0.14) (0.11) (0.39) (0.08) (101.53)

Household Income 100–199% FPL 0.92 –1.67** 0.78*** 1.24** 1.07 1.66** 0.67** –861.87***

(0.07) (0.69) (0.06) (0.11) (0.08) (0.35) (0.11) (97.44)

Household Income 200–399% FPL reference reference reference reference reference reference reference reference

Household Income 400%+ FPL 2.03*** 5.05*** 1.66*** 1.13 1.74*** 0.87 0.88 2,025.16***

(0.17) (0.70) (0.13) (0.10) (0.14) (0.27) (0.14) (126.86)

Mother Respondent Health Is Fair/Poor 0.89 –0.91 0.85* 1.07 0.89 1.31 0.80 –243.19***

(0.08) (0.81) (0.08) (0.11) (0.08) (0.23) (0.19) (93.84)

Parental Education Level High School 0.89 –0.01 0.98 0.88 0.75*** 2.13*** 1.36 –1,115.75***

(0.08) (0.76) (0.09) (0.08) (0.06) (0.58) (0.27) (113.61)

Parental Education Level Some College 1.18** 1.77*** 1.19** 0.97 0.92 2.31*** 1.14 –881.80***

(0.09) (0.65) (0.09) (0.08) (0.07) (0.58) (0.18) (104.72)

Parental Education College or Higher reference reference reference reference reference reference reference reference

Urban City, Urban Large reference reference reference reference reference reference reference reference

Urban, Small 1.21** 1.91** 1.19** 0.98 0.96 1.60** 2.67*** –532.84***

(0.11) (0.76) (0.11) (0.09) (0.08) (0.30) (0.41) (100.02)

Rural 1.15* 0.88 1.19** 0.95 1.00 2.13*** 1.60** –799.70***

(0.10) (0.72) (0.10) (0.09) (0.08) (0.38) (0.29) (98.02)

Constant 1.62 19.85*** 0.47** 0.43** 0.53** 0.00*** 0.18*** 4,507.85***

(0.50) (2.85) (0.16) (0.14) (0.17) (0.00) (0.10) (475.68) Observations 6,600 6,600 6,600 6,600 6,600 6,500 6,600 5,350

*** p<0.01, ** p<0.05, * p<0.1Standard errors in parentheses. Models also include state fixed effects. Regular care indicates a child was in nonparental care at least ten hours each week; PT indicates part-time care (between 10-29 hours/week); FT indicates full-time nonparental care (30 hours/week or more). aResults presented as odds ratios for logistic regression model bOLS regression models Source: Authors’ estimates using the ECLS-B


54

Appendix D: Regression Models Predicting Employment Outcomes

*** p<0.01, ** p<0.05, * p<0.1 Standard errors in parentheses. FT employment indicates the household mother was employed at least 35 hours/week; PT employment indicates household mother was employed less than 35 hours/week. aResults presented as marginal effects for logistic regression model bOLS regression models Source: Authors’ calculations using ECLS-B

Wave 3 and Kindergarten Enrollment by Disability Status

Wave 3 Kindergarten

Em

ployed a

Em

ployed F

Ta

Em

ployed P

Ta

Hours

Worked

b

Total H

H

Hours

b

Father

Em

ployeda

Em

ployed a

Em

ployed F

Ta

Em

ployed P

Ta

Hours

Worked

b

Total H

H

hoursb

Father

Em

ployeda

Ful

l Sam

ple Any Disability, Before Kindergarten –0.01 –0.02* 0.01 –0.93* –1.62** –0.02** –0.03** –0.02* –0.00 –1.11** –1.52** -0.01

(0.01) (0.01) (0.01) (0.54) (0.66) (0.01) (0.01) (0.01) (0.01) (0.54) (0.69) (0.01)

Continuing Disability, Before Kindergarten 0.00 0.00 0.00 –0.85 –1.14 –0.01 –0.03* –0.03 –0.00 –1.81** –2.53** -0.02*

(0.02) (0.02) (0.02) (0.80) (0.98) (0.01) (0.02) (0.02) (0.02) (0.79) (1.07) (0.01)

< 1

00%

FP

L

Any Disability, Before Kindergarten –0.03 –0.03 0.00 –1.02 –1.60 –0.02 –0.03 –0.02 –0.01 –0.97 –1.34 -0.01

(0.02) (0.02) (0.02) (0.88) (1.13) (0.02) (0.02) (0.02) (0.02) (0.93) (1.21) (0.02)

Continuing Disability, Before Kindergarten –0.04 –0.04 0.00 –2.00* –1.46 0.01 –0.03* –0.04 –0.00 –3.11** –2.55 -0.01

(0.03) (0.03) (0.02) (1.21) (1.55) (0.03) (0.02) (0.03) (0.02) (1.25) (1.82) (0.03)

SSI/

DI

Any Disability, Before Kindergarten –0.02 0.03 0.01 –0.59 –0.63 0.03 –0.03 –0.04 0.02 –3.13* 0.08 0.08

(0.04) (0.03) (0.01) (1.61) (2.09) (0.05) (0.04) (0.04) (0.03) (1.65) (2.10) (0.05)

Continuing Disability, Before Kindergarten 0.00 0.02 –0.01 0.21 1.85 0.06 –0.03* –0.06 0.01 –4.67*** –1.17 0.05

(0.04) (0.04) (0.03) (1.79) (2.30) (0.06) (0.02) (0.05) (0.03) (1.78) (2.36) (0.06)


55

child care for families raising children with disabilities ... · in the cost of college tuition...

Documents